Obstacle avoidance using VO

[jyotishp]

Objective

Make a single agent avoid multiple obstacles while reaching the goal.

What to do?

• Add a function inSensorRange that will take the state of the agent and a single obstacle and return true if the obstacle is within the sensor range of the agent.
• Add a function getConstraints that returns the non-linear constraints accepted by fmincon using VO.
• When writing VO constraints, do not copy-paste the code from somewhere. Try to write the constraints in your own terms. Once you get this working, I'll share how everyone in the lab typically does it.
• This function should be able to take multiple obstacles as input (array of obstacles) and should return the corresponding constraint values.
• The obstacle should be passed to the getConstraints only when the inSensorRange for that obstacle is true.
• Initially test it for single agent - single obstacle case. Then proceed to multiple obstacles case.
• Generate videos for {1, 2, 4, 6, 10} obstacles. The obstacles can overlap/collide but not with the agent.

[susiejojo]

For the constraints, we need to take the complement of the set of velocities lying in the collision cone right? So technically, the complement of VO should be the returned constraints?

[jyotishp]

No. Constraint can either be an equality or an inequality.

[jyotishp]

Check the examples on https://www.mathworks.com/help/optim/ug/fmincon.html

[susiejojo]

Yeah inequality only, but it should be such that the velocity doesn't fall into VO cone right? So the opposite inequality of what we needed to check if a velocity lies inside the cone?