Marmot Lab

280 subscribers

45 videos

Welcome to the Multi-Agent Robotic Motion (MARMOT) Laboratory at the National University of Singapore (NUS)! Lab Director: Prof Guillaume Sartoretti NUS Webp...

Latest Videos

[CoRL25] Search-TTA A Multimodal Test-Time Adaptation Framework for Visual Search in the Wild

[RSS25] SATA Safe and Adaptive Torque-Based Locomotion Policies Inspired by Animal Learning

[IROS24] Implicit Rendezvous for Robotic Exploration Teams under Sparse Intermittent Connectivity

ViPER Visibility-based Pursuit-Evasion via Reinforcement Learning - Spotlight Video (CoRL 2024)

Decaying Action Priors for Accelerated Imitation Learning of Torque-Based Legged Locomotion Policies

[Favorable Scenario] - ForMIC - 64 Agents, Infinite Resources, 0% Obstacles, 3 Long Wipeouts

[Favorable Scenario] - Centralized - 64 Agents, Infinite Resources, 0% Obstacles, 3 Long Wipeouts

[Average Scenario] - Centralized - 32 Agents, Depleting Resources, 0% Obstacles, No Wipeouts

[Unfavorable Scenario] - ForMIC - 32 Agents, Depleting Resources, 5% Obstacles, No Wipeouts

[Favorable Scenario] - Cardinality-MR - 64 Agents, Infinite Resources, 0% Obstacles, 3 Long Wipeouts

[Favorable Scenario] - C-SAF-11 - 64 Agents, Infinite Resources, 0% Obstacles, 3 Long Wipeouts

[Unfavorable Scenario] - Centralized - 32 Agents, Depleting Resources, 5% Obstacles, No Wipeouts

[Average Scenario] - C-SAF-11 - 32 Agents, Depleting Resources, 0% Obstacles, No Wipeouts

[Average Scenario] - Cardinality-MR - 32 Agents, Depleting Resources, 0% Obstacles, No Wipeouts

[Unfavorable Scenario] - C-SAF-11 - 32 Agents, Depleting Resources, 5% Obstacles, No Wipeouts

[Average Scenario] - ForMIC - 32 Agents, Depleting Resources, 0% Obstacles, No Wipeouts

[Unfavorable Scenario] - Cardinality-MR - 32 Agents, Depleting Resources, 5% Obstacles, No Wipeouts

Yixuan XIA - First Fourier-Based Footfall Planning Success

[No-Gradient Pheromones Test] - ForMIC - 64 Agents, Infinite Resources, 10% Obstacles, No Wipeouts

DAN Decentralized Attention-based Neural Network to Solve the MinMax mTSP

ForMIC Foraging via Multiagent RL with Implicit Communication

Pushing a wheeled cart using a hind legged gait

Gait transitions between common hexapedal gaits

Alternate tripod to hind legged gait transition

Gait transition to hind legged gait to enable mobile manupulation

[Noisy Pheromones Test] - ForMIC - 16 Agents, Infinite Resources, 0% Obstacles, No Wipeouts

PRIMAL2 Pathfinding via Reinforcement and Imitation Multi-Agent Learning - Lifelong

Decentralised Multi Agent Path Finding with Heterogeneous Speeds

Model based Dynamic Obstacle Avoidance on Inclined Surface

Obstacle Avoidance for A Hexapod Robot Based on FFT control