Topic

Reward Shaping

The design of reward functions that steer a reinforcement learning agent toward intended behaviour while closing off shortcuts that score well but look wrong.

Current clusters

build1 publisher

One GPU and a bash retry queue trained 14 skills for a bipedal robot duck

MicroDuck's 14 policies came out of roughly five days of serial single-GPU training in mjlab, with nine separate reward designs behind them and a VRAM ceiling that cut the env count to 2048 on rough terrain.

Publishers:dev.to

Reality

Evidence45
Adoption20
Hype gap+15
Incentives25
Confidence55