Session: ASME Undergraduate Student Design Expo

Paper Number: 174560

Development of a High-Accuracy Simulation for the Swerve Holonomic System 

Simulation tools for robotic systems are crucial in the development process, especially for complex systems such as swerve drives. Applications of such a simulation include the development of new control algorithms prior to testing on-device, or where such testing is not possible, reducing research time and cost. However, work on an accurate simulation for the swerve holonomic system is severely limited. Some simulation tools built with Robot Operating System (ROS) and Gazebo are capable of accurately modeling constraints present in certain holonomic systems, but fail to model the nonlinear dynamics inherent to a motor. Furthermore, these systems cannot model current limits or external loads, which are almost always present on real-world systems. To overcome these limitations, new mathematically-based solutions are required to conduct simulations with given constraints and those inherent to the system. The objective of this work is to design a simulation that closely replicates the behavior and dynamics of a real-world swerve system. Swerve drives are unique in their ability to control the orientation and speed of each wheel independently. This capability enables movement in any direction, even while the chassis is rotating. Applications for these systems might include navigating tight spaces and where precise lateral alignment might be needed, such as warehouses or manufacturing plants, or for inspection and maintenance. The simulation system developed in this work utilizes a 100 Hz discrete control loop, which simulates the system behavior at two levels. Level one includes a motor simulation, which reflects changes in velocity and position due to an individual duty-cycle control target, through the use of derived governing equations of motion. The motor-level simulation takes into account motor specifications, friction, system inertia, current limits, and other system properties. Level two incorporates a chassis simulation, which leverages the motor-level simulation to determine changes in its state as a result of an overall chassis control target. The chassis-level simulation first converts the linear and angular velocity targets into individual module (wheel and associated motors) targets, through a geometrical and vector math approach. The individual motor targets are then fed into the motor-level simulation to determine the changes in position and velocity of each module. The changes in module state are furthermore used to update the state of the chassis following an algorithm to update the information in an automated fashion. The end-to-end simulation is capable of closely modeling real-world dynamics. Future work may further the capabilities of this simulation by including the effect of wheel slip and tipping due to high acceleration. Our poster will include full details on the mathematical model and equations that were used in this simulation. Simulation results will be presented for various chassis targets to comprehend the simulation’s application for swerve holonomic systems.

Presenting Author: Vyaas Baskar Intelliscience Institute

Presenting Author Biography: I am working as a research intern at Intelliscience Institute and I am also affiliated to Dr. Zaidi's research group at San Jose State University. My interests include robotics and control systems. I am deeply involved in projects investigating various techniques to improve control of robotic arms in medical applications.

Authors:

Vyaas Baskar Intelliscience Institute
Sohail Zaidi San Jose State University