EAI Project

Install

1. Execute pip install -e ./ManiSkill
2. Install Vulkan, refer to https://autodl.com/docs/vulkan/ if using AutoDL.

Experiments

Below are the success rates of each benchmark task in simulation:

Task	Success Rate
Lift	$81.7%$
Stack	$66.7%$
Sort	$79%$
Arrange (Random instruction)	$83.3%$

See more details in the report.

The trajectories are stored in THU cloud disk. All trajectories are collected without privileged information, which is implemented by setting obs_mode=rgb so that obs_mode_struct.use_state=False in the ManiSkill environments.

Videos

Some testing videos of evaluation are in the videos folder.

Code Structures

Environments

• Environments of benchmark tasks are in so101_lift_cube.py, so101_stack_cube.py and so101_sort_cube.py.

• The file so101_lift_cube_v2.py is only for the so101_arrange family to inhere from.

• so101_arrange.py, so101_arrange_secondary.py are for training, while so101_arrange_eval.py and so101_arrange_color.py are for evaluation.

Training

You can use *_ppo.py for training. The code is modified from ppo_rgb.py in examples of ManiSkill, adding a bag of tricks including tanh squashing, state running average and std normalization, and LayerNorm before rgb last feature output.

Control

Inside grasp_cube/agents/robots/so101 there are so_101_ee.py and so_101_ee_new_rest_qpos.py for end-effector control and different rest positions.

Evaluating for task Arrange

For the default environment eval_arrange.py, initialized cube colors are R G B in order, and your input is the desired final order.

---

self.cube_perm_idx[env_idx] = torch.randint(low=0, high=1, size=(b,), device=device) # Change to 6 if random initial color permutation

Modify line 200 in so101_arrange_eval.py with high=6 to test random initial state. (Note that for easier controlling parallel environments, the sequence of swaps are still designed for RGB->Input, same for the following)

---

Modify import so101_arrange_eval to import so101_arrange_color and env_id: str = "ArrangeCubeSO101Eval-v0" to env_id: str = "ArrangeCubeSO101Color-v0" in eval_arrange.py, to test the generalization of the task with different colors.

Task and Methodology Highlights

The self-defined task, Arrange, introduces several sources of difficulty:

• Long horizon: The task contains multiple substeps of lifting and placing cubes. Our methodrequires 75 seconds in average if the given permutation is random, with a maximum of 150seconds to finish the task in simulation time.

• Dual arm: Both arms occupy overlapping spatial regions, and can not reach the regionfurthest from them. (For the example in the figure, the left arm can not reach the blue cube.)So two-arm coordination is required for this task.

• Instruction-related: The task requires an input as the instruction, which is the desired finalconfiguration for the robot to execute.

• Generalization: Our method can be generalized to the task with different color of cubes. (See arrange_different_color.mp4 in videos)