top of page
Continual Learning for Imitation Learning
- Data Aggregation
- Continual Learning
Data-driven AD, Planning
2025
- DAgger for Imitation Learning
- Continuous closed-loop simulation
- Identify failure / long-tail cases where the ML planner cannot handle the situation
- Training with Failure & Long-tail Data
- Accumulate the refined data and perform continual fine-tuning for sustained performance gains
bottom of page