CN102666030B - A method for reducing the energy consumption of an industrial robot and an industrial robot system - Google Patents

Abstract

The present invention relates to a method for reducing the energy consumption of an industrial robot comprising a manipulator (8) having a plurality of arms that are movable relative each other about a plurality of axes. The method comprises: defining a model for the energy consumption of the robot dependent on the movements of the axes of the robot, including a relation between the energy consumed due to friction and the speed of the axes, and the energy consumed due to gravity acting on the arms considering the fact that the energy consumed due to gravity can be reduced during standstill one or more of the axes if the axes are mechanically locked, storing a control program specifying a geometric path to be followed by the robot, determining, during at least a part of the work cycle, speed profiles for the axes of the robot when following the specified geometric path with regard to minimizing the energy consumption of the robot and determining whether or not an axis should be mechanically locked during standstill in order to reduce the energy consumption, based on said model for the energy consumption of the robot, and a maximum allowed time for carrying out the robot movement during said part of the work cycle, provided that it is permitted to complete the geometric path in a shorter time than the maximum allowed time, and calculating reference values for the motors of the robot based on the determined speed profiles.

Description

For reducing method and the industrial robot system of the energy ezpenditure of industrial robot

Technical field

The present invention relates to a kind of method for reducing the energy ezpenditure being configured to the industrial robot performing work during the work period.In addition, the present invention relates to a kind of industrial robot system comprising robot, described robot comprises executor and for storing the program storage of the control program of work that definition will be performed by robot during the work period.

Background technology

Industrial robot is usually used to industrial automation.Robot is programmed to follow mobile route while robot performs work during the work period.Robot control program has should do about robot for what and the information should executed the task along this path according to which order.Control program comprises a series of program instruction of writing with robot programming language.Control program has the information about all impact points on mobile route.Robotic programming can carry out manually online completing by the position of the impact point of professor robot on mobile route and orientation.Can also by the offline programming run on the outer computer of such as PC and simulation tool to produce control program.

Industrial robot system comprises executor, and it has and relative to each other around multiple arm of multiple axle movement and the motor for activating the movement of axle, and can have the robot controller for manipulation device.Robot controller comprises the memory cell of the one or more control program for storing the movement for controlling executor.Control program comprises programmed instruction, and this programmed instruction comprises the move for executor.The geometric path will followed when control program specifies in robot execution work.This program can also be specified along the speed in path and the restriction of acceleration.This geometric path can be represented with joint angle or cartesian coordinate.Relation between joint angle and cartesian coordinate is determined by the motion model of robot.The position of this geometric path designated robot instead of speed.

Robot controller comprises and is configured to executive control program and the program interpreter providing instruction based on move to path planner.Path planner receives instruction from program interpreter and determines how executor moves can perform move based on this.Path planner generates the track comprising geometric path and the VELOCITY DISTRIBUTION along this path.Path planner plans how to perform indicated movement by the interpolation performing movement.This interpolation comprises indicated movement is divided into multiple little increment, and calculates the joint angle for all axles for robot of each increment.This joint angle is converted into a reference value for motor subsequently.The motor benchmark of calculating is sent to driver element by path planner, and this driver element controls motor by converting DC electric current to variable AC current according to this motor benchmark.

An important parameter for nearly all robot application is cycle time.Mean robot cycle time and perform time of spending work period.Control program is importantly made to be time optimal to reduce cycle time as far as possible.Therefore, normal route planning algorithm is the solution based on optimization problem, and its cycle time being defined as making to stand client constraints and robot constraint minimizes, and wherein, client constraints can be the speed and accekeration of such as being specified by user.Some example of robot constraint is maximum gearbox torque, maximum available motor torque and maximum motor speed.This means in each discrete time step, use all available motor power, only otherwise violate client or robot retrains.

Now, in nearly all branch of commercial Application, industrial robot is used everywhere.Especially as advantage, robot such as contributes the total environment impact of factory to some extent due to its energy ezpenditure.The energy ezpenditure of work period depends on the movement of robot, such as the Actual path of robot during the work period, speed and acceleration.Therefore, there is the object reducing the energy that robot consumes.

The paper of Andre R Hirakawa and Atsuo Kawamura in minutes the 3rd volume, 1626-1632 page of 1996IEEE application meeting on October 6th, 1996 " Trajectory generation for redundant manipulators under optimizationof consumed electrical energy " discloses a kind of method of the track energy optimization for making to have set time length for the robot with redundant degree of freedom.D1 utilizes self-movement the in inside caused due to redundancy motion carry out energy optimization and make to minimize with the deviation of given cartesian trajectories.A shortcoming of the method only in certain degree of accuracy, follows geometric path.In addition, if be applied to nonredundancy robot, then want optimized unique be freely the deviation with cartesian trajectories.Follow in the very important application of given geometric path wherein, this method will have the possibility of very little minimizing energy ezpenditure.

EP 1705541 display can be carried out time optimization to the time-critical part of work period and during the non-time critical part of work period, reduce the speed of robot, to reduce energy ezpenditure.During the non-time critical part of work period, robot advanced with alap speed according to the maximum permission time period determined.But reduction speed not necessarily reduces the energy consumed due to friction and gravity.

Summary of the invention

The object of the invention is to reduce further the energy that robot consumes when following and specifying geometric path.

According to an aspect of the present invention, by the method defined in claim 1 to realize this object.

These class methods comprise:

-model of the energy ezpenditure for robot is defined according to the movement of the axle of robot, comprise due to the friction of axle and speed and relation between the energy consumed and the energy consumed due to the gravity acted on arm, mechanically locked if to consider in axle one or more, the energy consumed due to gravity can be reduced during one or more transfixion then in executor or its axle

-store the control program of geometric path that designated robot will follow when the work of execution,

-the work period at least partially period, based on the described model of the energy ezpenditure for robot and the maximum permission time for performing robot movement during the described part of work period, about making the minimum energy consumption determination robot of robot VELOCITY DISTRIBUTION for the axle of robot when following appointment geometric path, and determine whether axle should be mechanically locked to reduce energy ezpenditure during transfixion, condition allows to complete geometric path within the time shorter than the maximum permission time, and

-a reference value of the motor for robot is calculated based on determined VELOCITY DISTRIBUTION.

According to the present invention, make robot in the VELOCITY DISTRIBUTION optimization following the axle for robot when specifying geometric path about making the minimum energy consumption of robot.VELOCITY DISTRIBUTION for axle comprises the information of speed about axle and the how time dependent information of speed and the acceleration about axle therefore.The present invention makes the speed of executor and acceleration instead of geometric path optimization.Geometric path is fixing, or can be allowed to change in little tolerance interval.The present invention is conducive to for following given path such as to avoid the application conflicted with the fixed obstacle in robot working space to depending on.

For mechanical system, power and torque are square proportional, and therefore rub and to play an important role to energy ezpenditure.Model for the energy ezpenditure of robot comprises the relation between energy and the speed of axle consumed due to friction.There is the speed that wherein friction torque is in minimum of a value.During speed when the speed of executor is in minimum of a value lower than friction torque, energy ezpenditure will increase, as shown in Figure 3.By using the model for energy ezpenditure considering the energy ezpenditure caused due to friction, the robot speed of minimises power consumption can be determined.As a result, the time for the movement of execution robot of robot can be shorter than the maximum permission time.This possibility is used to reduce the energy consumed further.According to the present invention, optimization is performed under permission completes the condition of geometric path within the time shorter than the maximum permission time, thus make it possible to be in its final position place in robot and locked by the axle of robot, and turn off the control of motor subsequently to save energy.

According to the present invention, the fact that the energy consumed due to gravity when one or more locked in axle is significantly reduced to be considered during the optimization of robot movement.This axle is such as mechanically locked by the brake of applied robot.Whether energy is saved the length depending on the actionless time period, application and release brake cost regular hour and preferably those times are also included in a model.If the actionless time period is too short, then impossible brake application device, or the consumption of brake application device is than the more energy saved.Whether optimal method determination axle should mechanically be locked to reduce energy ezpenditure during transfixion.Axle mechanically can be locked during the transfixion of any time during the energy optimization part of work period, is not only when mobile end.Preferably, the impact on energy ezpenditure when considering axle to lock during transfixion all the time.This means to reduce energy ezpenditure further by gathering way for some situation.Therefore, it is possible to carry out the further minimizing of energy ezpenditure.

The invention enables can about making the minimum energy consumption of robot and the optimization of maximum permission time order in the movement of the robot of period at least partially of work period for performing work.Use term during the period at least partially of work period means the one or more parts during the whole work period or in the work period.The present invention is suitable at least partially period not crucial robot application of time in the work period, such as wherein robot just at tend machinery and robot the part of work period is spent in the application waited on machine readiness.The invention enables and can save energy for this type of application.Arrange the maximum permission time for performing work according to application, such as robot must wait for that machine is by the minimum time looked after.

According to embodiments of the invention, the model for energy ezpenditure comprises the time that its mechanical lock engaging/depart from the axle of executor spends, and it is designed to consider the time for the engagement/disengaging of mechanical lock.Whether energy is saved the length depending on the actionless time period, apply and discharge brake to spend the regular hour and preferably those times are also included in a model.If the actionless time period is too short, then impossible brake application device, or the consumption of brake application device is than the more energy saved.

According to embodiments of the invention, the movement of executor is determined about the minimum energy consumption made during the part of work period, and about the movement making the robot time minimum that execution work spends during another part of work period determine executor, and based on about making the minimum energy consumption of robot and moving a reference value of the motor calculated for robot about the robot that the time minimum made needed for robot execution work is determined.The present embodiment makes it possible to cycle time is minimized and is making to switch between the minimum energy consumption during the work period.The present embodiment is favourable to the application during the work period with one or more time-critical part and one or more non-time critical part.

According to embodiments of the invention, control program comprises the instruction of the minimum energy consumption for turning on and off robot during the work period.The present embodiment makes robot program person can the minimizing of energy ezpenditure during the selected portion in command job cycle.

According to embodiments of the invention, the method comprises and calculates by about making the minimum energy consumption of robot instead of about making minimize cycle time to determine the mobile energy realized of executor to reduce and showing calculated energy minimizing.The present embodiment makes robot operator can watch realized energy minimizing and encourage robot operator to use this function hopefully.

According to embodiments of the invention, described robot comprises driver element, it is by converting DC electric current to variable AC current control motor according to being used for the described a reference value of motor, and comprises the model of the model of the energy ezpenditure of the mechanical part of control device, the model of the energy ezpenditure of motor and the energy ezpenditure of driver element for the described model of the energy ezpenditure of robot.

According to a further aspect in the invention, by the industrial robot system defined in claim 8 to realize this object.

Described robot system comprises:

-data storage, it stores the model depending on the energy ezpenditure for robot of the movement of the axle of robot, described model comprises the relation between the energy that consumes and the energy consumed due to the gravity acted on arm due to friction and the speed of axle, and described model is designed to consider such fact, if one or more namely in axle are mechanically locked, then can reduce the gravity owing to acting on arm during the transfixion of executor and the energy consumed, and

-energy optimization module, its be configured to for the work period at least partially follow specify geometric path time and about the VELOCITY DISTRIBUTION making the minimum energy consumption of robot determine the axle of robot, and based on described control program, determine whether axle should be mechanically locked to reduce energy ezpenditure during transfixion for the described model of the energy ezpenditure of robot and the maximum permission time for performing robot movement during the described part of work period, condition allows to complete geometric path within the time shorter than the maximum permission time, and

-computing unit, it is configured to a reference value moving the motor calculated for robot based on determined robot.

According to embodiments of the invention, this system comprises time optimization module, it is configured to the movement determining executor at least partially based on the dynamic model of robot about the time minimum making robot execution work spend for the work period, this system to be configured to when receiving orders at the time minimum making robot execution work spend and to make to switch between the minimum energy consumption of robot during a work period, and described computing unit is configured to based on about making the minimum energy consumption of robot and moving a reference value of the motor calculated for robot about the robot that the time minimum making robot execution work spend is determined.

According to embodiments of the invention, control program comprises the instruction switched between time minimum for making robot execution work spend and the minimum energy consumption making robot during the work period, and described system is configured to switch between time and energy optimization based on the instruction in robot program.

According to embodiments of the invention, this system comprises the outer computer for the offline programming of robot and the robot controller for the movement that controls executor, described energy optimization module is configured to generate energy optimization control program and this energy optimization module is stored on outer computer, and the computing unit for a reference value calculating the motor for robot is stored in robot controller.The present embodiment makes it possible to about minimises power consumption during the offline programming of robot and makes control program optimization.Output from offline programming is energy optimization control program.The advantage of the present embodiment is easily realization and there is not the demand of computational speed, and therefore can use the optimization algorithm of more demands with high accuracy.

According to embodiments of the invention, this system comprises robot controller, it comprises the path planner being suitable for determining executor how movement based on control program, and described energy optimization module and computing unit are a part for path planner and are stored on robot controller.According to the present embodiment, during path planning, in robot controller, complete the energy optimization of movement.If robot program is not offline programming, then from the angle of user on the controller instead of perform optimization at outer computer and have superiority.Another advantage is the maximum permission time can be variable, and such as it can depend on the input signal from external unit.Another advantage of the present embodiment is because path planner to determine the fact in path based on little increment, if compared with performing energy optimization with off line, and can with more high accuracy determination track.

Accompanying drawing explanation

Body more closely will explain the present invention with reference to accompanying drawing with the description of different embodiments of the invention now.

Fig. 1 illustrates for the time optimization VELOCITY DISTRIBUTION of manipulator shaft and in the energy optimization VELOCITY DISTRIBUTION not considering to determine when rubbing.

Fig. 2 illustrate Fig. 1 time optimization VELOCITY DISTRIBUTION and about the energy optimization VELOCITY DISTRIBUTION determined of rubbing.

Fig. 3 illustrates the example of the relation between friction torque and speed.

Fig. 4 illustrates the example of prior art robot controller.

Fig. 5 illustrates robot system according to an embodiment of the invention.

Fig. 6 illustrates the example of the energy optimization completed on outer computer.

Fig. 7 illustrates to comprise and only makes VELOCITY DISTRIBUTION and do not make the optimized energy optimization of geometric path estimate the robot controller generated.

Fig. 8 illustrates that robot controller generation energy optimization track comprises and determines energy optimization geometric path and energy optimization VELOCITY DISTRIBUTION.

Fig. 9 illustrates the example of the assembly in robot system energy model.

Figure 10 illustrates the simplification robot with an only rotary joint.

Detailed description of the invention

The present invention minimizes based on consumption of energy instead of cycle time, and this can describe with following optimization problem:

Minimize: the integration of energy ezpenditure

Obey: client constraints

Robot retrains

Cycle time≤the maximum cycle time

Wherein, MaxCycleTime (maximum cycle time) defines the maximum cycle time for the user of robot movement.Client constraints can be the speed and accekeration of such as being specified by user.Robot constraint is such as maximum gearbox torque, maximum available motor torque and maximum motor speed.

Exist many wherein robots be not bottleneck and time wherein between robot motion by according to such as robot the machine looked after and predefined and wherein use maximum time instead of the logical robot application of minimum time.

The solution of optimization problem will be discussed in detail after a while, but now by some general aspects proposing to separate and the algorithm that will find solution.

Term energy optimization is used as the equivalent of energy the best or energy optimization motion planning hereinafter, this means that planning robot moves so that minimises power consumption.

Energy optimization can be switched on and turn off.When energy optimization is turned off, carry out motion planning about making to minimize cycle time.According to embodiments of the invention, the turning on and off, such as, by being provided for the programmed instruction turned on and off of energy optimization of order energy optimization in a control program.Can each program, each module, each process or each instruction connection energy optimization.

It is below the example how using energy optimization in the program code of control program.

Energy optimization is ordered EnergyOptimizationOn to open and is closed by order EnergyOptimizationOff.Two instruction EnergyOptimizationOn and EnergyOptimizationOff are included in move sequence.EnergyOptimizationOn connects energy optimization and the second independent variable allows motion before performing EnergyOptimizationOff instruction, be allowed to the maximum time spent.Programmed instruction between EnergyOptimizationOn and EnergyOptimizationOff instruction is called as energy optimization block.The time given by the variable maxMoveTime in this example can become as the function of other correlated process, and therefore, it can be different when each execution of particular energy optimization block that actual energy is saved.Note that once control program performs entered this block, then can not change the largest motion time parameter of energy optimization block.The calculating that the energy that order printEnergyReduction initiates to be realized by energy optimization instead of the optimization of cycle time reduces, and the energy that display calculates reduces.

From some comment to this solution of the angle of user: if too short for the maximum time of energy optimization block, then this means that normal minimum period time standard may can not find the track meeting the designated period time, program performs and will stop with error message, such as " time of specifying for energy optimization is too short ".There is not the track with < cycle time " xx second ".Replace " xx " with performing the time that in fact move spend according to optimal standards.If have selected the time close to " xx " in EnergyoptimizationOn instruction, then do not leave the too many free degree for energy optimization, and therefore reduce realizing little energy.When performing optimization, calculate the estimation that energy reduces, and if to the given optional independent variable " printEnergyReduction " of EnergyoptimizationOff order, then this value is printed on such as teaching unit.

The pith of optimization concept makes it possible to lock to the axle application machine of executor on final position, the brake of such as robot, and then turn off the control of motor to save energy.When reaching the time of specifying in EnergyoptimizationOn instruction, brake must be released, and motor must be in controlled mode, can respond, and may next instruction after energy optimization block time directly movement.

In the standard industrial robot's configuration be minimized cycle time wherein, the geometric path that robot should follow is defined well by from robot control program.In order to carry out energy optimization, path can be left and come given by robot program and make VELOCITY DISTRIBUTION optimization, or making geometric path be an optimized part.This causes two different situations in optimization:

Situation 1: geometric path is given and only have dynamics may be affected (this means the VELOCITY DISTRIBUTION along path) in optimization.

Situation 2: only have and start target and target end and to be designated and geometric path is freely selected in optimization.Situation 2 is not a part of the present invention.

But, comprise wherein situation 1 and situation 2 are combined the situation making geometric path given be possible and useful, wherein, robot its reduce total power consumption time may from path deviation to certain prescribed limit.

In the following example, the object code of the maximum allowable offset had from the 100mm expecting geometric path is shown:

Optimization algorithm can be performed in different levels,

Level 1: based on model in off line instrument (such as exterior PC), and before program is run on robot controller.

Level 2: based on model, but in robot controller:

A. robot control program start before precalculate.

B. control program the term of execution online calculating.

The different types of solution will be described in detail after a while.This solution is similar in level 1 and 2a.Essential difference between varying level is the requirement to computational complexity.In level 1 and to a certain extent in horizontal 2a, there are not the many restrictions to computational complexity, and horizontal 2b has very strict restriction to computational complexity because must while robot movement and with in robot controller can limited computer resources come optimization.

In energy optimization block, exercise data is not allowed to change when performing this block.Do not allow sentence of having ready conditions, i.e. if statement or recursion instruction, it may affect path.When entering energy optimization block, the geometry in the path in necessary decision block.

Use and be based on the important difference between the optimized level 1 of model and level 2: in the execution at level 1 place, can not Update Table when performing a programme, and when being in level 2 place, exercise data can change, until enter energy optimization block.

Fig. 1 illustrates time optimization VELOCITY DISTRIBUTION 1 for the manipulator shaft of the movement when not being affected by gravity and energy optimization VELOCITY DISTRIBUTION 2.Energy optimization VELOCITY DISTRIBUTION makes robot reach home in appointment maximum time exactly.

Fig. 3 illustrates the typical frictional behaviour with static friction, viscous, coulomb and viscous friction.For mechanical system, the power produced and torque are square proportional, therefore rub to play an important role for the energy ezpenditure of robot.In figure 3, show typical frictional behaviour, and clearly, there is the speed that wherein friction torque is in minimum of a value, therefore, if realize energy optimization about friction, then compared to Figure 1 will cause different VELOCITY DISTRIBUTION.

The time optimization VELOCITY DISTRIBUTION that Fig. 2 illustrates Fig. 1 and the energy optimization VELOCITY DISTRIBUTION 3 determined about friction.The energy optimization VELOCITY DISTRIBUTION that result obtains provides acceleration higher compared with the distribution shown in Fig. 1 and speed, but acceleration lower compared with distributing time optimal and speed.Energy optimization VELOCITY DISTRIBUTION 3 illustrates that the time point Tr before the Max time arrives the motion of final position.In the time of the time from Tr to Max, mechanical brake may be activated, therefore interval room machine system not consumed energy at this moment, and this reduces total power consumption further.In such as Fig. 1, the energy regenerated when robot slows down at the end of moving in most of the cases is wasted as the electric energy by resistor (producing the heat energy of a great deal of).Energetic optimum is dissolved also must consider this energy and the minimum of kinetic energy making waste, and in fact, this will mean and underspeeds, and cause the solution in Fig. 2.Therefore the speed that obtains of result by be make the minimum of kinetic energy of waste and make frictional dissipation minimize between balance.

In order to reduce energy ezpenditure further, if one or more axle is being long enough to apply and be actionless during the time period discharging brake, then can also during robot motion to one or more axle brake application device.

Can the optimal standards of track time optimal be used for by assessment and calculate total energy compared with it being realized with in energy optimum trajectory and reduce.Show total energy to reduce.

Fig. 4 illustrates the example of prior art robot system, and it comprises, and have can relative to each other around multiple arm of multiple axle movement with for the executor 8 of motor that activates the movement of axle and the robot controller 9 for the movement that controls executor.Robot controller 9 comprises for storing definition by the program storage 10 of one or more control programs of the work performed during the work period by robot and the program interpreter 12 being configured to executive control program.

Robot controller 9 also comprises the path planner from program interpreter reception instruction and determines how executor moves can perform move based on this.Path planner generates the track comprising geometric path and the VELOCITY DISTRIBUTION along this path.Path planner comprises geometric path planner 14, to geometric path (namely it be configured to, the position in path) plan, and dynamic optimization planner 16 is configured to plan VELOCITY DISTRIBUTION for path about making minimize cycle time.Dynamic optimization planner 16 uses the algorithm based on the solution to optimization problem, and the cycle time that described optimization problem is defined as making to stand client constraints and robot constraint minimizes.Controller 9 also comprises the data storage 18,19,20 of storing machine people motion model, Robotic Dynamic model and drive system model.According to control program, robot motion model is used to produce geometric path.Dynamic optimization planner uses geometric path to use the dynamic model of robot and the model of drive system of robot and the distribution of generation time optimum speed.

Path planner also comprises computing unit 21, and it is configured to a reference value calculating the motor for robot based on the path determined and VELOCITY DISTRIBUTION.Computing unit 21 performs the interpolation of the track generated.This interpolation comprises track is divided into multiple little increment, and calculates the joint angle for all axles for robot of each increment.This joint angle is converted into a reference value for motor subsequently.The motor benchmark of calculating is sent to the drive system 22 of robot controller by computing unit 21.Drive system comprises by converting DC electric current to driver element that variable AC current controls motor according to motor benchmark.

Fig. 5 illustrates the example according to robot system of the present invention.Introduce new additional conditions the Track Pick-up in Conventional temporal optimization situation and new energy optimization situation to be differentiated.In Figure 5, show how must revise overall structure to introduce energy optimization Track Pick-up concurrently with current time optimization Track Pick-up.Program interpreter detected energy optimization is opened by order or is closed.The energy optimization order EnergyOptimizationOn be such as closed in a control program opens and is closed by order EnergyOptimizationOff.Energy optimization can be unlocked and close repeatedly during the work period.If energy optimization is closed, then for movement instruction time of implementation optimization Track Pick-up 30.Move the time optimal that time optimization Track Pick-up 30 determines robot and comprise and generate geometric path based on robot motion model and the step generating VELOCITY DISTRIBUTION time optimal based on Robotic Dynamic model and drive system model.Movement instruction to time optimized input.

If energy optimization is unlocked, then start energy optimization Track Pick-up 30.Programmed instruction (also referred to as program block) between EnergyOptimizationOn and EnergyOptimizationOff instruction performs energy optimization.It is program block and the maximum time for performing this block to the input of energy optimization.Based on all movement instructions in program block with perform energy optimization for the maximum time of this block.Energy optimum trajectory generates the energy optimization determining robot and moves and comprise and generate geometric path based on robot motion model and the step generating the distribution of energy optimum speed based on robot system energy model, Robotic Dynamic model and drive system model.Based on such as moving for performing the energy optimization that maximum permission time of the above-mentioned maxMoveTime of movement and robot and usual constraint generates for robot.According to one embodiment of present invention, energy optimization geometric path and energy optimization VELOCITY DISTRIBUTION is generated.

Described robot system comprises the time optimization module 30 being configured to time of implementation optimization Track Pick-up and the energy optimization module 32 being configured to perform energy optimization Track Pick-up.Robot system comprises the data storage 34 for storing machine robot system energy model.Robot system energy model is the model of the energy ezpenditure of the robot system of the movement of the axle depending on robot.Robot system comprises robot controller, and it comprises the computing unit 21 being configured to calculate a reference value of the motor for robot based on determined movement.

Robot about the minimum energy consumption making robot moves the movement that optimization means to determine robot for the purpose of minimises power consumption.How accurately have according to energy model, the energy ezpenditure that can realize in various degree reduces.

Can run by the varying level robot system from the visible energy optimization of introducing of the above-mentioned energy optimization problem followed it.Therefore, it is possible to generate and the generation of energy optimum trajectory in robot controller or at the term of execution deadline optimum trajectory of control program with off-line mode or during the programming of robot on outer computer during the programming of robot.No matter be complete energy optimization on the controller or on outer computer, on robot controller, complete the calculating of motor benchmark all the time.

In figure 6, show level 1, i.e. off line energy optimization.At outer computer 40, place completes energy optimization offline, such as, during the offline programming of robot.Input under this level is robot control program, and output is the new energy optimization control program comprising energy optimization ro-bot movement instruction.For the programmed instruction that wherein energy optimization is closed, programmed instruction is not modified.If energy optimization is opened, then perform energy optimization on the programmed instruction (also referred to as program block) between EnergyOptimizationOn and EnergyOptimizationOff instruction.Output from energy optimization is energy optimization program block.Program block comprises move, and energy optimization program block comprises energy optimization programmed instruction.In addition, the instruction of engagement/disengagement brake can be inserted in a program.

In optimization, utilize robot motion model, the model of dynamic model and drive system and other robot system energy model.If use the situation 1 of energy optimization, energy optimization program can have the path geometry structure identical with original program, if or the combination of service condition 2 or situation 1 and situation 2, then there is the path geometry structure identical with amendment geometry.At the end of programming, control program is loaded into robot controller.Off line energy optimization module is gone forward side by side using robot program as input the energy optimization of a part of line program, and wherein, energy optimization is activated.Output from energy optimization module is a part for new energy optimization robot program or program.Energy optimization program in robot controller the term of execution, path planner generates motor benchmark based on the move in control program.

Fig. 7 illustrates robot controller 50, and it comprises only makes VELOCITY DISTRIBUTION and brake engagement/disengagement moment instead of the optimized energy optimization of geometric path.In this case, the geometry of ro-bot movement instruction's designated robot motion of control program, makes VELOCITY DISTRIBUTION and brake engagement/disengagement moment optimization to realize minimal energy consumption simultaneously.Path planner comprises geometric path planner 14, and it generates geometric path based on the move in control program and robot motion model.Path planner also comprises energy optimization module 52, and it is configured to generate energy optimization VELOCITY DISTRIBUTION based on the model of geometric path description, Robotic Dynamic model, drive system and robot system energy model.Path planner also comprises time optimization module 53, and it is configured to generate time optimization VELOCITY DISTRIBUTION based on Robotic Dynamic model and drive system model.If energy optimization is unlocked, then energy optimization module 52 is used to formation speed distribution and brake engagement/disengagement moment, if energy optimization is closed simultaneously, then time optimization module is used to formation speed distribution.

Fig. 8 illustrates the robot controller 54 comprising and be suitable for the path planner 56 generating energy optimization track.The generation of energy optimization track comprises determines energy optimization geometric path and energy optimization VELOCITY DISTRIBUTION.In this case, the geometry of freely decision paths in energy optimization.Robot program provides and starts and terminate robot configuration, and next, energy trajectory optimization finds energy optimal path for robot and VELOCITY DISTRIBUTION.Optimum use robot motion model, dynamic model, drive system model and last robot system energy model obtain energy optimum trajectory.

When optimization is run for 2 times in level, situation 1 and the situation 2 of principle of optimality can be described as in figures 7 and 8.In case 1, still robot motion model is used to generate geometry, simultaneously with utilizing the energy optimization of Robotic Dynamic model, drive system model and robot system energy model to replace time optimization by geometric path planner.The benchmark that the motor benchmark that result obtains is reserved as the motor of robotic manipulator 8 is fed to drive system 22 downwards.In situation 2, Fig. 8, geometry is also the free parameter that energy optimization will judge.This means that geometric path planner is a part for energy optimization block 56 now and the input to this block is robot program, output is the benchmark to drive system.Energy optimization utilizes robot motion model, Robotic Dynamic model, drive system model and last robot system energy model in situation 2.

This energy model is the model of the energy ezpenditure of complete machine robot system; This comprises manipulator, motor and power drive system.Other resource in system can also be taken into account, all if be switched on according to the current state of system and turn off or carry out the fan that controls and processor.In fig .9, the block diagram for energy model is shown.This model comprises mechanical model, power drive system model, motor model and controller hardware model.In addition, this model comprises environmental model, such as can measures ambient temperature with accomplishing the input that friction gauge is calculated.

Mechanical model comprises the Robotic Dynamic model comprising friction model.But the mechanical model in energy model can have the friction model more accurate than prior art mechanical model.Power drive system model and motor model are the expansions of the drive system model of energy flow (re-using of energy) between motor when can comprise acceleration/deceleration.Energy dissipation model is also included in motor model, and this model comprises the actual efficiency of friction in motor and motor.Controller hardware model is the model of the energy ezpenditure in the such as controller hardware such as fan, CPU board.This model also comprises the model of brake; The engagement/disengagement time and engagement and clastotype in energy ezpenditure.

Two joints will be divided to provide detailed description.In first segment, the simplified version of problem is solved to introduce the concept of solution and explains this solution depends on anything.In second section, introduce identical problem and describe for complete, actual robot solution to model.

simplification situation

Figure 10 illustrates the diagram in simplification robot and electricity and machinery two fields with an only rotary joint.Also illustrate the use of optional arm, select with this, example becomes identical with the motion of (such as the axle 3 of hinged 6-DOF robot) of robot.

In this case, use the simplification system according to Figure 10, wherein, describe mechanical part with following formula

$J\stackrel{\&CenterDot;\&CenterDot;}{q}+d\left(q\right)+f\left(\stackrel{\&CenterDot;}{q}\right)=\mathrm{\&tau;}---\left(1\right)$

Wherein, J is inertia (depending on matrix or the scalar of the number of the free degree of corresponding system), and q is arm angle.Gravity item when d (q) represents that arm is attached to load, as shown in Figure 10. be friction torque and τ represents the actuator torque produced by such as motor.With following DC motor equation, electric part is described:

$L\stackrel{\&CenterDot;}{i}+\mathrm{Ri}+u_m=u$

Wherein, i is circuital current, and u is input voltage, and L is motor inductance, and R is motor resistance and u _mit is the back-EMF (motor back emf) generated according to following equation;

τ＝ki， $u_m=k\stackrel{\&CenterDot;}{q}$

Wherein, k is torque constant.In addition, should add dissipation item in energy model, usually, this is nonlinear function h (i, the u of electric current and back-EMF _m).

Friction can be described as

1) only viscous friction, $f\left(\stackrel{\&CenterDot;}{q}\right)=f_m\stackrel{\&CenterDot;}{q}$

2) there is the model of Cologne friction and viscous friction, such as by Canudas de Wit, C, Olsson, H. and Lischinsky K., P. (1995) are at IEEE Transactionson Automatic Control, the Lugre model proposed in " the A new model for control ofsystems with friction " of 40 (3): 419-425, or by B.Feeny and F.Moon. (1994) at Journal of Sound and Vibration, propose in " the Chaosin a forced dry-friction oscillator:Experiments and numericalmodelling " of 170 (3): 303-323 that.

$f\left(\stackrel{\&CenterDot;}{q}\right)=f_m\stackrel{\&CenterDot;}{q}+f_c({\mathrm{\&mu;}}_k+(1-{\mathrm{\&mu;}}_k){\cosh }^{-1}\left(\mathrm{\&beta;}\stackrel{\&CenterDot;}{q}\right))\tanh \left(\mathrm{\&alpha;}\stackrel{\&CenterDot;}{q}\right)$

According to the dynamic (dynamical) equation of descriptive system, optimization problem can be stated as now

min∫Pdt

$\stackrel{\&CenterDot;}{x}=f(x,u)$

s.t.

u∈U (2)

Wherein, x is the state of dynamical system, and u is the input to system, and f is nonlinear function, and P can be divided into electric P and mechanical P.Electricity P is by the gross energy of system consumption and mechanical P is only mechanical part.In this case in order to can energy model be clearly stated, again, the equation describing motor model is write:

$u=L\stackrel{\&CenterDot;}{i}+\mathrm{Ri}+u_m+h(i,u_m)$

And electric motor drive system model:

$u=K_p(i_{\mathrm{ref}}-i)+K_I\&Integral;(i_{\mathrm{ref}}-i)\mathrm{dt}=K_p(\frac{\mathrm{\&tau;}}{k}-i)+K_I\&Integral;(\frac{\mathrm{\&tau;}}{k}-i)\mathrm{dt}$

Wherein, K _pand K _icurrent controller ratio and storage gain.Simply, current reference is the torque t calculated by following formula:

$J\stackrel{\&CenterDot;\&CenterDot;}{q}+d\left(q\right)+f\left(\stackrel{\&CenterDot;}{q}\right)=\mathrm{\&tau;}$

Suppose that current controller is fast speed, make i follow i well _ref.Finally, energy model can be formulated as now

E＝∫Pdt＝∫uidt

$u=L\stackrel{\&CenterDot;}{i}+\mathrm{Ri}+u_m+h(i,u_m)$

Motor

$u=K_p(\frac{\mathrm{\&tau;}}{k}-i)+K_I\&Integral;(\frac{\mathrm{\&tau;}}{k}-i)\mathrm{dt}$

Electric motor drive system

${\mathrm{\&tau;}}_i=\left\{\begin{array}{ccc}{\mathrm{Ja}}_i+d\left(q\left(t_i\right)\right)+f\left(\stackrel{\&CenterDot;}{q}\left(t_i\right)\right)& \mathrm{if}& \stackrel{\&CenterDot;}{q}\left(t_i\right)\&NotEqual;0\\ 0& \mathrm{if}& \stackrel{\&CenterDot;}{q}\left(t_i\right)=0\end{array}\right.$

Mechanical model

τ＝ki， $u_m=k\stackrel{\&CenterDot;}{q}$

Electricity is to machine power conversion

Input to model is joint acceleration a _i.Can by simple integration from a _icalculate joint velocity and position.In above energy model, do not comprise the model of controller hardware energy ezpenditure.The brake engagement/disengagement time is assumed to be zero here.

When motor is in stable state, when namely i is constant, provide electric energy by P=ui, wherein i=T/k and u=Ri, namely clearly, static torque level during robot transfixion has a significant impact energy ezpenditure tool, because it occurs with quadratic form.Therefore, clearly, wherein the actionless position of robot is important for total power consumption.By brake application device, waste energy during robot transfixion can be reduced.The mechanical model that Here it is why in above energy model has two different situations, one be when joint is moved and one be when joint transfixion.But this model is simplified, because it is instantaneous that brake engagement/disengagement is assumed to be.

In order to above-mentioned general duty Optimization, can utilize discrete versions, wherein, q (t) is parameterized, makes acceleration be free parameter in each interval, and then can be by speed and position calculation:

$\stackrel{\&CenterDot;}{q}(t_i+{\mathrm{\&Delta;}}_t)=\stackrel{\&CenterDot;}{q}\left(t_i\right)+a_i{\mathrm{\&Delta;}}_t---\left(3a\right)$

$q(t_i+{\mathrm{\&Delta;}}_t)=q\left(t_i\right)+\stackrel{\&CenterDot;}{q}\left(t_i\right){\mathrm{\&Delta;}}_t+\frac{a_i{\mathrm{\&Delta;}}_t^2}{2}---\left(3b\right)$

Wherein Δ _tcan be fixing or from calculate and a _iat time interval t _ito t _i+1the accekeration of middle use.In order to simplify this solution, also supposing that current controller is obviously faster than optimization step-length, namely can incite somebody to action in each optimization moment be assumed to be zero.Now this optimization problem is formulated as:

$\underset{a}{\min }\underset{i=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{u}_i\&CenterDot;{\mathrm{\&tau;}}_i$

s.t.

$u_i=R\frac{{\mathrm{\&tau;}}_i}{k}+k\stackrel{\&CenterDot;}{q}\left(t_i\right)+\tilde{h}({\mathrm{\&tau;}}_i,\stackrel{\&CenterDot;}{q}\left(t_i\right))$

${\mathrm{\&tau;}}_i=\left\{\begin{array}{ccc}{\mathrm{Ja}}_i+d\left(q\left(t_i\right)\right)+f\left(\stackrel{\&CenterDot;}{q}\left(t_i\right)\right)& \mathrm{if}& \stackrel{\&CenterDot;}{q}\left(t_i\right)\&NotEqual;0\\ 0& \mathrm{if}& \stackrel{\&CenterDot;}{q}\left(t_i\right)=0\end{array}\right.$

τ _min≤τ _i≤τ _max

a _min≤a _i≤a _max

t _N＝T _max

q(t ₀)＝q ₀

$\stackrel{\&CenterDot;}{q}\left(t_0\right)={\stackrel{\&CenterDot;}{q}}_0$

q(t _N)＝q _N

$\stackrel{\&CenterDot;}{q}\left(t_N\right)={\stackrel{\&CenterDot;}{q}}_N$

${\stackrel{\&CenterDot;}{q}}_{\min }\&le;\stackrel{\&CenterDot;}{q}\left(t_i\right)\&le;{\stackrel{\&CenterDot;}{q}}_{\max }---\left(4\right)$

In (4), with such as carrying out formulism to many constraints of torque, joint velocity and joint acceleration to discrete optimization problem.Use τ=ki, formulism is carried out again with can with mathematical formulae to write optimization problem to the standard in energy model and equation.Current controller is used obviously to suppose motor model and electric motor drive system to combine faster than optimization time step.Torque is provided by the system equation in (1).This constraint comprises about the maximum of torque and acceleration and minimum of a value and the boundary condition about Angle Position and angular speed.Normal condition will be that angular speed is at time started (t ₀) be zero and at end time (t _n) be zero, but the constraint (start, terminate and maximum/minimum of a value) of other value and angular acceleration can also be comprised.

Suppose that incremental time is fixed in sampling time Δ _tand this value is less, separate more closely close to true optimum value continuously.Formula in given (4), Standard General nonlinear optimization algorithm can be used to come duty Optimization, see the Numerical Optimization of such as Jorge Nocedal and StephenJ.Wright, Springer Series in OperationsResearch, the PracticalMethods of Optimization.1987.ISBN 0-471-91547-5 of 1999, ISBN 0-387-98793-2 and R.Fletcher.In (4), suppose the brake application device when robots arm's transfixion, because time do not apply torque.

the solution developed completely

When developing completely, carry out optimization relative to the motion of energy ezpenditure to the robot with two or more frees degree (DOF).Optimization can be carried out, that is, according to previously described situation

1. given geometric path and only have dynamics to be affected in optimization,

2. only specify and start target and target end and freely select geometric path in optimization.

Comprise wherein 1 and 2 to be combined, make to provide wherein robot its reduce total power consumption situation can from path deviation to the situation of the geometric path of certain prescribed limit be possible and useful.

Can to carry out formulism with mode similar in (4) to two of this problem kinds of situations.In situation 1, the geometry of given path, and (scalar) VELOCITY DISTRIBUTION that optimization problem will find along the energy and braking engagement/disengagement moment minimized geometry being used in each axle.In situation 2, except beginning and end position, do not specify geometry.This situation is similar to (4), wherein, only specifies and starts and end point and acceleration is free parameter.In the situation 2 of the solution developed completely, a _iparameter is no longer scalar, alternatively, and each a _iit is the vector of the length of the number of the actuating free degree with the robot equaling considered.Still can to carry out formulism with mode similar in (4) to optimization problem, but some inequality becomes vector inequality now.The complexity that the number of variable is conciliate also increases significantly along with the number of degrees of freedom increased.

The invention is not restricted to the disclosed embodiments, but can carry out changing and revising in following right.Such as, energy optimization can be opened and closed with the signal of the external unit from such as PLC.

Claims (12)

1. one kind for reducing the method for the energy ezpenditure of industrial robot, described industrial robot is configured to execution work during the work period, and comprise that have can relative to each other around multiple arm of multiple axle movement and the executor of motor for activating the movement of axle, wherein said method comprises:

-model of the energy ezpenditure for robot is defined according to the movement of the axle of robot, comprise the speed due to friction and axle and relation between the energy that consumes and the energy consumed due to the gravity acted on arm,

-store the control program of geometric path that designated robot will follow when the work of execution, and

-based on the described model of the energy ezpenditure for robot, the robot about the period at least partially making the minimum energy consumption of robot determine in the work period moves, and

-a reference value of the motor calculated for robot is moved based on determined robot, it is characterized in that,

-be designed to consider such fact for the described model of energy ezpenditure, if namely axle is mechanically locked, then can reduce the gravity acted on arm during the one or more transfixion in axle, and

-describedly determine that the robot of the period at least partially in the work period moves and comprises: based on the described model of the energy ezpenditure for robot and for the described part in the work period during perform maximum permission time of robot movement, calculate the VELOCITY DISTRIBUTION of the axle for following robot when specifying geometric path, and determine whether axle should mechanically be locked to reduce energy ezpenditure during the transfixion of axle, condition allows to complete geometric path within the time shorter than the maximum permission time.

2. method according to claim 1, the described model wherein for energy ezpenditure is designed to the engagement/disengagement time being used for mechanical lock to take into account.

3. method according to claim 1 and 2, is wherein locked shaft mechanical by the motor applications brake to robot.

4. method according to claim 1, the movement of executor is wherein determined about the minimum energy consumption made during the part of work period, and about the movement making the robot time minimum that execution work spends during another part of work period determine executor, and based on about making the minimum energy consumption of robot and moving a reference value of the motor calculated for robot about the robot that the time minimum made needed for robot execution work is determined.

5. method according to claim 1, wherein said control program comprises the instruction of the minimum energy consumption for turning on and off robot during the work period.

6. method according to claim 1, wherein said method comprises:

-calculate by about the minimum energy consumption making robot, instead of about making minimize cycle time, determine that the energy that the movement of executor realizes reduces, and

-showing the energy calculated reduces.

7. method according to claim 1, wherein robot comprises driver element, described driver element by converting DC electric current to variable AC current control motor according to being used for the described a reference value of motor, and comprises the model of the model of the energy ezpenditure of the mechanical part of executor, the model of the energy ezpenditure of motor and the energy ezpenditure of driver element for the described model of the energy ezpenditure of robot.

8. one kind comprises the industrial robot system of robot, described robot comprises that have can relative to each other around multiple arm of multiple axle movement and the executor (8) of motor for activating the movement of axle, and for storing the program storage (10) of the control program of geometric path specifying in robot when robot performs work during the work period and will follow

-data storage (34), it stores the model depending on the energy ezpenditure for robot of the movement of the axle of robot, described model comprises the relation between the energy that consumes and the energy consumed due to the gravity acted on arm due to friction and the speed of axle, and

-energy optimization module (32), its be configured to for the work period at least partially, based on described control program, for the described model of the energy ezpenditure of robot about the movement making the minimum energy consumption of robot determine executor, and

-computing unit (21), it is configured to a reference value moving the motor calculated for robot based on determined robot,

It is characterized in that,

Described data storage is suitable for storing the model for energy, described model is designed to consider such fact, if namely axle is mechanically locked, then during one or more in axle transfixions, the gravity owing to acting on arm can be reduced and the energy consumed, and

-described energy optimization module is configured to based on the described model of the energy ezpenditure for robot and the maximum permission time for performing robot movement during the described part of work period, calculating robot follow specify geometric path time for the VELOCITY DISTRIBUTION of the axle of robot, and determine whether axle should mechanically be locked during transfixion, to reduce energy ezpenditure, condition allows to complete geometric path within the time shorter than the maximum permission time.

9. robot system according to claim 8, wherein said system comprises time optimization module (30), its be configured to for the work period at least partially, determine the movement of executor about the time minimum making robot execution work spend based on the dynamic model of robot, described system to be configured to when receiving orders at the time minimum making robot execution work spend and to make to switch between the minimum energy consumption of robot during a work period, and described computing unit is configured to based on about making the minimum energy consumption of robot and moving about the robot that the time minimum making robot execution work spend is determined, calculate a reference value of the motor being used for robot.

10. robot system according to claim 9, wherein said control program comprises the instruction for switching between the time minimum making robot execution work spend and the minimum energy consumption making robot during the work period, and described system is configured to switch between time and energy optimization based on the instruction in robot program.

11. robot systems described in any one according to Claim 8 in-10, wherein said system comprises the outer computer (40) of the offline programming for robot and the robot controller (50 for the movement that controls executor, 54), described energy optimization module (32) is configured to generate energy optimization control program, described energy optimization mould is certainly stored on outer computer, and computing unit (21) is stored on robot controller.

12. robot systems according to claim 8, wherein said system comprises robot controller (50,54), it comprises the path planner being suitable for determining executor how movement based on control program, and described energy optimization module (32) and computing unit (21) are parts for path planner.