MimicGen produces large-scale datasets with minimal human effort

MimicGen can generate diverse datasets from 10 human demos including...

New Reset Distributions

In the example below, MimicGen generated 1000 demos for each of 3 reset distributions from just 10 human demos.

We showcase several datasets generated by MimicGen across broad task reset distributions below.

New Objects

New Robot Hardware

Long-Horizon Tasks

High-Precision Tasks

Mobile Manipulation

Real-World Tasks

MimicGen datasets can produce performant policies across diverse tasks with simple Behavioral Cloning

Additional Results and Visualizations

MimicGen can produce good quality datasets and policies across different quality human operators.

Below, we show MimicGen datasets generated on Square D2 from two sets of source datasets - 10 demos from a better quality operator and 10 demos from a worse quality operator (both are from the robomimic Square MH dataset). Surprisingly, policies trained on each dataset achieve comparable results, which suggests that in the large-scale data regime, data quality might not matter as much.

Using MimicGen to generate equal amounts of data as a human operator (e.g. 200 demos generated from 10 human vs. 200 human demos) can result in comparable policy performance.

Below, we show policies trained on 200 demos on Square D0 -- on the left, the policy was trained on 200 demos generated by MimicGen from 10 human demos, and on the right, the policy was trained on 200 human demos. The agent performance is comparable. This raises important questions about the presence of redundancies in large human datasets and when to request additional data from a human.

MimicGen Data Generation Example

Tasks and Reset Distributions

Tasks

Reset Distributions

MimicGen can be robust to noisy pose estimates