thanard / hallucinative-topological-memory

Hallucinative Topological Memory

Hallucinative Topological Memory (HTM) tackles the problem of Visual Planning in which an agent learns to plan goal-directed behavior from observations of a dynamical system obtained offline, e.g., images obtained from self-supervised robot interaction. In particular, HTM learns an energy-based model based on contrastive loss and a conditional VAE model that generates samples given a context image of a new domain. It uses these hallucinated samples for nodes, and energy-based model for the connectivity to build a planning graph. HTM allows for zero-shot generalization to domain changes.

The environment currently consists of a Block Wall domain (see https://arxiv.org/abs/2002.12336) made easy for testing zero-shot generalization.

Set-Up

1. Install standard ML libraries through pip/conda and Mujoco.
2. Change file to execution mode by chmod +x scripts/collect-data.sh
3. Run ./scripts/collect-data.sh to collect data, or download the training and test data here.

Training

1. Change to execution mode and run ./scripts/train_vae.sh.
2. Change to execution mode and run ./scripts/train_actor.sh.
3. Change to execution mode and run ./scripts/train_cpc.sh.

Evaluation

1. Change mode and run ./scripts/evaluate.sh