Doctoral Thesis Proposal - Benjamin Stoler

— 2:00pm

Location:
In Person and Virtual - ET - Reddy Conference Room, Gates Hillman 4405 and Zoom

Speaker:
BENJAMIN STOLER, Ph.D. Student, Computer Science Department, Carnegie Mellon University
https://benstoler.com/

Towards Robust Autonomous Driving and Social Robot Navigation via Enhanced Data Utilization

Autonomous robots—including self-driving vehicles, sidewalk delivery robots, and more—must navigate among humans in a safe and socially-compliant manner. Current approaches for building and evaluating such autonomous systems rely on data-driven techniques; however, a generalization gap emerges, as methods trained in these traditional paradigms are unable to cope with unexpected real-world scenarios. Therefore, this thesis aims to develop improved evaluation settings and methodologies to increase and assess robustness in autonomous robot navigation against these challenges. This thesis proposal describes several completed works that assess and improve different facets of robustness in autonomy:

  • For robustness against perception errors affecting downstream motion prediction, we construct a framework for converting top-down pedestrian trajectory datasets into a more challenging first-person view perspective. We then develop a correction module to account for the resulting errors, trained end-to-end with trajectory prediction approaches.
  • For robustness against out-of-distribution, safety-relevant scenarios, we create a hierarchical characterization method which leverages counterfactual probes to find hidden safety-relevant scenarios in large datasets. We then address the induced generalization gap by incorporating the characterizations into downstream trajectory prediction models' inductive biases.
  • For robustness against adversarial, safety-critical scenarios, we develop a reactive, skill-based adversary policy which leverages a learned, multi-faceted criticality objective to perturb existing scenarios. We then train ego policies in a closed-loop manner against these generated scenarios, demonstrating improved downstream ego performance.

This proposal concludes by outlining and discussing proposed works to further advance robustness in autonomous navigation. These works include enhancements in scenario characterization and out-of-distribution generalization, as well as novel formulations of realism as an objective in safety-critical generation. 

Thesis Committee

Jean Oh (Chair)
Sebastian Scherer
Reid Simmons
Jonathan Francis (Bosch Center for Artificial Intelligence)
 

Additional Information

In Person and Zoom Participation.  See announcement.

Event Website:
https://csd.cmu.edu/calendar/doctoral-thesis-proposal-benjamin-stoler