← Work

Category:

Robotics

Role:

Author

Years:

2021 to present

Field:

Reinforcement learning

Autonomous navigation in an unknown environment

Media

Autonomous navigation in an unknown environment, building the airframe
building the airframe
navigation test run
state-space diagram

Notes

A mobile robot that learns to navigate a space it has never seen, using a hybrid of DQN and REINFORCE to get the sample efficiency of one and the continuous-action behaviour of the other.

Started as my master's project, presented at ICECET 2022 in Prague, now extended into a 21-page manuscript under review at IEEE Transactions on Robotics.