avanturist2025MIT

MIKASA-Robo

A comprehensive benchmark suite of memory-intensive robotic manipulation tasks with ready-made datasets and pre-trained oracle agent checkpoints for reinforcement learning research.

Downloads1K
Likes4

Why This Matters for Physical AI

MIKASA-Robo provides a benchmark for training and evaluating reinforcement learning agents on memory-intensive robotic manipulation tasks, advancing research in solving complex real-world robotic problems that require temporal reasoning and state tracking.

Technical Profile

Environment
simulationtabletop
Task Types
manipulationmemory-intensive tasks
License
MIT
Part of the MIKASA-Robo family

Community Signals

Top 25% by downloads

Access

Need custom physical AI data?

Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.

Request a Sample Pack