ropiens / project-sandwich-man

A project for researching a complex and long-horizon manipulation task especially focused on hierarchically stacking blocks.
MIT License
5 stars 0 forks source link

[S1-01-2] PandaPush, PandaStack, assign @CUN-bjy #5

Closed CUN-bjy closed 3 years ago

CUN-bjy commented 3 years ago

PandaPush-v1

Screenshot from 2021-08-24 10-58-22 Peek 2021-08-24 10-59

benthebear93 commented 3 years ago

PandaStack-v1

PandaStack_log PandaStack

CUN-bjy commented 3 years ago

It seems impossible to learn a single-step stacking task with a sparse reward(e.g. PandaStack). We need to apply the multi-step(hierarchical) structure!