Search papers, labs, and topics across Lattice.
DreamerNav is introduced, a robot-agnostic navigation framework built upon DreamerV3, that tackles autonomous navigation in dynamic indoor environments by integrating multimodal spatial perception, hybrid global-local planning, and curriculum-based training. The system uses egocentric depth images and a local occupancy map to represent dynamic obstacles and integrates this with a global A* path within a POMDP framework. Results demonstrate superior success rates and adaptability compared to NoMaD, ViNT, and A* in simulation and successful zero-shot transfer to real-world quadrupedal robots.
DreamerNav achieves robust zero-shot transfer of autonomous navigation policies learned in simulation to real-world quadrupedal robots navigating dynamic indoor environments.
Robust autonomous navigation in complex, dynamic indoor environments remains a central challenge in robotics, requiring agents to make adaptive decisions in real time under partial observability and uncertain obstacle motion. This paper presents DreamerNav, a robot-agnostic navigation framework that extends DreamerV3, a state-of-the-art world model–based reinforcement learning algorithm, with multimodal spatial perception, hybrid global–local planning, and curriculum-based training. By formulating navigation as a Partially Observable Markov Decision Process (POMDP), the system enables agents to integrate egocentric depth images with a structured local occupancy map encoding dynamic obstacle positions, historical trajectories, points of interest, and a global A* path. A Recurrent State-Space Model (RSSM) learns stochastic and deterministic latent dynamics, supporting long-horizon prediction and collision-free path planning in cluttered, dynamic scenes. Training is carried out in high-fidelity, photorealistic simulation using NVIDIA Isaac Sim, gradually increasing task complexity to improve learning stability, sample efficiency, and generalization. We benchmark against NoMaD, ViNT, and A*, showing superior success rates and adaptability in dynamic environments. Real-world proof-of-concept trials on two quadrupedal robots without retraining further validate the framework’s robustness and quadruped robot platform independence.
