Human Aware Motion Planner Using Augmented PRM

This project is in the proposal/proof-of-concept stage in my lab. Our goal is to improve the human experience in collaborative/close-proximity tasks by using a human-aware motion planner.

My approach to this is two-fold:

1. Generate a PRM, augmented with information to make human-aware calculations cheaper, such as:
   1. Collision probabilities
   2. Values needed for kinematic and dynamic calculations
2. Find paths in the graph, using a modified A* search, but calculating human-aware costs, such as:
   1. Probability of collision
   2. Proximity to human
   3. Rapid motions towards human

Demonstration PRM

As the initial proof-of-concept (and easy-to-debug example), I’ve created a 2D world, with:

• An obstacle (black square)
• high-probability collision regions (grey)

There are several attempts at planing from the upper-left to lower-right corners:

• Red: standard A*
• Pink: cost for being above the downward diagonal (right-shifted)
• Brown: cost for being high up in the world
• Grey: cost for moving directly towards the black object
• Blue: cost for moving into high-probability-collision nodes (only planner with this knowledge)

Upcoming Work

1. Extension to robot arm planning
2. Improvements to PRM generation
3. Investigating properly adapting A* search to these additional costs