Active learning from demonstration for robust autonomous navigation

By David Silver, J. Bagnell, Anthony Stentz. 2012

65 citations recorded by the source corpus.

Building robust and reliable autonomous navigation systems that generalize across environments and operating scenarios remains a core challenge in robotics. Machine learning has proven a significant aid in this task; in recent years learning from demonstration has become especially popular, leading to improved systems while requiring less expert tuning and interaction. However, these approaches still place a burden on the expert, specifically to choose the best demonstrations to provide. This work proposes two approaches for active learning from demonstration, in which the learning system requests specific demonstrations from the expert. The approaches identify examples for which expert demonstration is predicted to provide useful information on concepts which are either novel or uncertain to the current system. Experimental results demonstrate both improved generalization performance and reduced expert interaction when using these approaches.

Subfields: Artificial Intelligence, Control and Systems Engineering, Meta & Continual Learning, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Sergey Levine
2. George Dimitri Konidaris
3. Jan Peters
4. Harish Chaandar Ravichandar
5. Sonia Chernova
6. Andrew Barto
7. Pieter Abbeel
8. Jens Kober

In Control and Systems Engineering

1. Harish Chaandar Ravichandar
2. Jan Peters
3. George Dimitri Konidaris
4. Sonia Chernova
5. Jens Kober
6. Sergey Levine
7. Andrew Barto
8. Athanasios S. Polydoros

In Meta & Continual Learning

1. George Dimitri Konidaris
2. Bruno Castro da Silva
3. Andrew Barto
4. Sonia Chernova
5. David Silver
6. Kamil Andrzej Ciosek
7. Tomas Lozano-Perez
8. Leslie Pack Kaelbling

In Robotics & Control

1. Sonia Chernova
2. Andrew Barto
3. Bruno Castro da Silva
4. Jens Kober
5. George Dimitri Konidaris
6. Aude Gemma Billard
7. Harish Chaandar Ravichandar
8. Athanasios S. Polydoros

Citations in the corpus

• Learning Parameterized Skills (2012, background): “Finally, taking advantage of human demonstrations might help biasing EISP towards tasks which an expert deems relevant, which suggests an integration with active learning from demonstration techniques [62].”
• Learning Interactively to Resolve Ambiguity in Reference Frame Selection (2020, background): “In navigation applications, novelty detection based on density estimation is used for requesting demonstrations in uncertain situations [11].”
• Challenges in Autonomous Vehicle Testing and Validation (2016, background): “In recent years, most robotics applications have turned to machine learning to do this [21, 22], because the complexities of the multi-dimensional optimization are such that manual effort is unlikely to yield desired levels of performance.”
• Recent Advances in Robot Learning from Demonstration (2020, methodology): “Since this early success, LfD-based autonomous navigation for ground vehicles has been attempted using IRL (117, 121, 168), interactive learning (169), active learning (170), adversarial learning (171), and end-to-end learning (40, 172).”