Search this site
Embedded Files
Arunim Joarder
  • Home
  • About
  • Research
  • Resume
  • Contact
Arunim Joarder
  • Home
  • About
  • Research
  • Resume
  • Contact
  • More
    • Home
    • About
    • Research
    • Resume
    • Contact

Multi-agent deep reinforcement learning for quadrupedal walking

The Idea

The goal was to make independent agents for each leg of a quadrupedal robot. The intention was to inject robustness into the system, making the overall robot less dependent on each leg, allowing complex manoeuvres with one leg holding a door open while the other legs make the robot walk. Similarly, this independence from individual legs would lead to robust behaviours even in case of leg failures.

EmailLinkedInGitHub
Google Sites
Report abuse
Page details
Page updated
Google Sites
Report abuse