CyC GraphPlan

CyC GraphPlan is a path planning and motion control solution complementary to CyC GraphSense. It uses optimization and predictive control methods for calculating a safe trajectory in the environment reconstructed by CyC GraphSense.

Given a global reference trajectory \(\mathbf{z}_{ref}^{\langle t-\infty, t+\infty \rangle}\), CyC GraphPlan calculates a safe trajectory and control signals for the robot, while avoiding the obstacles found along this reference path.

The figure below shows CyC GraphPlan controlling an aerial vehicle along a reference trajectory (green) and the block diagram of the planning and control system. The method uses CyC GraphSense and the underlying robot state \(\mathbf{x}^{\langle t \rangle}\) to generate feasible trajectory candidates in the 3D Cartesian space.

Each candidate trajectory is evaluated using a cost function that penalizes obstacle collisions and the distance between each candidate waypoint and the given global reference path. The optimization loop simultaneously calculates the best (desired) trajectory \(\mathbf{z}_{d}^{\langle t+1, t+N \rangle}\) and the control signals \(\mathbf{u}^{\langle t \rangle}\) which would execute the robot's motion along \(\mathbf{z}_{d}^{\langle t+1, t+N \rangle}\). \([t+1, t+N]\) is the time interval for which the generated trajectories are calculated, encoding their length.

CyC GraphPlan is able to create paths in 3D space that avoid obstacles not only sideways. The quadrotor's path goes both sideways, as well as over the obstacles.

The videos showcase CyC GraphPlan both on Autonomous Ground Vehicles (AGV), where the planning is performed in 2D, as well as for controlling Aerial Vehicles (Drones), which involves motion control optimizations in 3D space.

Path planning and motion control in Autonomous Ground Vehicles (AGV)

Path planning and motion control in Aerial Vehicles (Drones)

Request a demo via CyC droids.