There are several aspects of the mechatronics challenge in robotics. In the previous post, I hinted at some, but on reflection the list is significant and all the issues should be considered.
Managing the trajectory is a challenge on computing level just to crunch the numbers and resolve the mechanical organization of the robot into the individual axes of motion. If we assume the move profiles are processed before a move is initiated, that task by itself pales in comparison to monitoring the individual axes on the fly and executing following window errors on all axes so that any mechanical error will bring the robot to a halt. Then, when the robot is halted, what exactly constitutes a “safe” condition? Hold position? Return to home at low speed? Everything needs to be well defined before a system can be put into operation.
Accuracy in a linear or Cartesian system is hard enough to deal with. When talking about 5,6 & 7 axis robots it is that much more difficult. It is not like a numerically controlled machine tool where all the axes are on solid tooling and there reference points and geometry of operation are relatively simple, even though very complex interactions can be programmed. It is a case where each axis contribute error in a tolerance stack up. So extra attention is typically paid to the mechanical integration of the robot itself to minimize these factors. Use of precision cycloidal gear reducers with high reduction ratios and low mass are one approach to the robot.
Force feedback from the any of the individual axes are handled through the torque (current) loop between the individual axis amplifier and motor. The problem gets complicated when moving through space. How much power is needed? This depends on how fast the robot needs to go. How much force is the needed to overcome the mass of the actuators themselves versus the force needed to move the load. This becomes a huge concern when human interaction with robots occur. The robot must not be able to exert forces or speeds that will injure a human. The human must be allowed to overcome the robot’s actuator system. So super lightweight systems must be designed with highly dynamic controllers to meet these special requirements.
Because of the unique nature of robot controllers, ntegration with other controls is often difficult. Robot controllers require stand-alone operation due to the critical nature of the how they operate. Integration with other control systems can be done with I/O handshaking or high speed communications, but it must be foolproof.
Lastly, integration with the environment an open ended aspect having more to do with the application than the basic technology. If the robot is going to perform the DARPA Automomous Vehicle Challenge, well, that is a whole different matter. But even on the factory floor, integration with the environment is not simple. Vision systems may be required to find the object and it’s orientation so the robot can acquire it. The speed and intelligence of the vision system now become key to the success of the application.
Robots continue stretch the limits and definitions of the things we call motion control.