Scientific researches in motor neuroscience reveal that the nervous systems of higher category animals are controlled by a modular approach.
This means that the complex movements of the muscles are constituted of basic motion primitives and the central nervous system or brain doesn’t care about every little detail of the movement.
But most of the industrial robot design haven’t adopted this principle of motion modularity. These robots have to generate complex trajectories with the help of numerous sampling joints and inverse kinematics.
This approach is similar to using the brain to directly govern and guide every specific movement of the muscle.
Researchers are trying to bring a revolutionary change in the future of robot control with motion modularity technology.
Let’s take a brief look at that.
Basis of Dynamic Movement Primitives (DMP)
Proposed by Professor Stefan Schaal of the University of Southern California, the Dynamic Movement Primitives (DMP) is a technique for robot trajectory control and planning.
This method can trace the trajectory better and also improves the generalization ability of the robot motion.
DMP finds applications in human motion modeling, human-robot collaboration, and the research of robot imitative learning. Scientists have also used this method to build a robot while advancing neurosurgical science research.
The primary idea of the DMP mechanism is to utilize a simple yet stable system to parameterize the expression of complicated trajectories and then adjust the entire system through a nonlinear function.
It inherits the benefits of linear systems, like robustness to interference, convergence to the attracting domain, and time independence. On the other hand, the inclusion of a force function lets the system generate random motion trajectories.
There are two major types of DMP models: rhythmic and discrete. Rhythmic systems are based on the limit cycles and discrete systems are based on attraction points.
Trajectory Generation Based on Motion Modularity Technology
The application of DMP is no more in the theoretical level. The roboticists used an actual robot design to analyze traditional industrial robots. SIA SmartPainting – a painting system for vehicle maintenance is used for this purpose.
Its controller is a usual centralized industrial robot control system. But according to the characteristics of the robot controlling task, the system can be divided into four levels: joint layer, communication layer, trajectory layer, and planning layer.
At the planning layer, the system can visualize the painted door to create a motion graph of the robot’s end-effector. Operations like inverse kinematics, trajectory interpolation, and time parameterization are performed at the trajectory layer.
Then the relevant data are transferred to the communication layer. This layer ensures the stability of the spray gun and the smoothness of the trajectory. Finally, the joint layer receives the reference signal and controls the joint motor motion according to that.
Conclusion
There’s been a lot of research on modular robots and we believe this will progress even further in the upcoming future. To find out more please contact us or call tel:+01.385.230.7377
2020