A collision cone represents all relative velocities that would cause two objects to collide within a given time. Think of it as a "danger zone" in velocity space.
Imagine two self-driving cars approaching an intersection. The collision cone shows all relative speeds and directions that would result in a crash. The autonomous system must choose velocities outside this cone.
Velocity Obstacles extend collision cones to handle moving obstacles. Instead of just relative velocity, we consider the actual velocities needed to avoid a moving obstacle.
A warehouse robot (blue) needs to avoid a moving cart (red):
When both agents are intelligent (like two robots), they should both contribute to collision avoidance. RVO splits the avoidance responsibility equally.
In a drone swarm, each drone uses RVO to avoid others:
Robots can't move at any velocity - they have physical and operational constraints that must be considered.
A robotic arm in a factory has multiple constraints:
Collision Cones: Show dangerous relative velocities in a geometric way
Velocity Obstacles: Extend cones to handle moving obstacles in real-time
RVO: Enables symmetric, oscillation-free multi-agent navigation
Constraints: Ensure solutions are physically realizable and operationally safe