Network Control Robots
simulate the control of 6 wheeled robots in PyBullet physics simulator.
Swarm Robotics
Swarming of robots refers to the collective behavior exhibited by a group of robots or agents that work together in a coordinated manner, often emulating the behavior seen in natural swarms such as bird flocks, fish schools, or insect colonies. Swarming involves individual robots following simple rules or behaviors that, when combined, result in emergent complex behaviors at the group level. This was a phd level course which I took in my masters in mechanical engineeirng, where at the end of the course we had to simulate six robots in pybullet engine to undergo controlled geometrical formation while avoiding the obstacles to reach the final destination.
Project Objective:
The goal is to simulate the control of 6 wheeled robots in PyBullet (a physics simulator) that need to navigate an environment. The robots should:
- Create a square formation in the original arena.
- Switch to a different formation to cross the narrow passages. In all formations, keep at least 1m between each robot.
- Navigate through the environment, avoid obstacles and reach the goal.
- Create a triangle formation once the goal has been reached.
Target
Description of the Controller:
A combination of the consensus algorithm and the rigidity-based control technique is employed to establish the formation and control of robots. To achieve the desired motion, a proportional-derivative control law is utilized to propel the formation forward. The equation are as shown below.
\[F_{\text{formation},i} = \begin{bmatrix} K_f \cdot (Dz_{(x, \text{desired})} - L_x)_i + D_f \cdot (\dot{z}_{(x,\text{desired})} - L_{\dot{x}})_i \\ K_f \cdot (Dz_{(y, \text{desired})} - L_y)_i + D_f \cdot (\dot{z}_{(y,\text{desired})} - L_{\dot{y}})_i \\ \end{bmatrix}\]Where πΎπ and π·π and gains, L is the graph Laplacian.
\[F_{target,i} = \begin{bmatrix} K_{T}*min(y_{\text{goal}} - y_i) + D_{T}\dot{x}_{i}\\ K_{T}*min(y_{\text{goal}} - y_i) + D_{T}\dot{y}_{i}\\ \end{bmatrix}\]Code
The python code for the project is on my Github repository.
Result
In the project, I effectively utilized the pybullet physics engine to simulate a cohesive swarm of robots. As a result, I successfully achieved the designated project objective.
