Materials: [ DualityObstacleAvoidancePolyhedra.pdf]
This video starts the discussion on how to incorporate collision avoidance into optimization-based approaches to trajectory optimization, to be addressed in several videos.
This is a continuation of the case study in videos [
In this video, we explore primal options in which no new decision variables are added relative to prior examples; i.e., the decision variables are essentially the robot waypoints and the control actions. Collision avoidance is encoded as a set of constraints, either with:
Spherical cover of robot and/or obstacles, so if a clearance larger than the radius of the sphere is guaranteed, then there will be no collision
A log-sum-exp smoothing , which gets closer to the actual obstacle boundary for large but produces a smooth contour out of the obstacle for low . The actual value of is application-dependent, considering the required obstacle clearance, accuracy, and solver time.
*Link to my [ whole collection] of videos in English. Link to larger [ Colección completa] in Spanish.