Prediction and Decision Making for Socially Interactive Autonomous Driving
Photos Taken on June 9
A photo of the presentation of Jiachen Li
Group photo of invited speakers and organizers
Prof. Mykel Kochenderfer, Stanford University
Mykel Kochenderfer is a professor of Aeronautics and Astronautics at Stanford University. He is the director of the Stanford Intelligent Systems Laboratory (SISL), conducting research on advanced algorithms and analytical methods for the design of robust decision making systems, ranging from unmanned aircraft to driverless cars. He is also the director of the SAIL-Toyota Center for AI Research at Stanford and a co-director of the Center for AI Safety. He received a Ph.D. from the University of Edinburgh and B.S. and M.S. degrees in computer science from Stanford University. He is an author of the textbooks "Decision Making under Uncertainty: Theory and Application" and "Algorithms for Optimization", both from MIT Press.
Prof. Katherine Driggs-Campbell, University of Illinois at Urbana-Champaign
Katie Driggs-Campbell is currently an Assistant Professor in the ECE Department at the University of Illinois at Urbana-Champaign. Her research focuses on exploring and uncovering structure in complex human-robot systems to create more intelligent, interactive autonomy. She draws from the fields of optimization, learning & AI, and control theory, applied to human robot interaction and autonomous vehicles. Previously, she was a postdoctoral research scholar in the Aero-Astro Department in the Stanford Intelligent Systems Lab and received her MS and PhD in Electrical Engineering and Computer Science from the University of California, Berkeley.
Jens Schulz , Technical University of Munich and BMW Group
Jens Schulz is a PhD student at the Chair of Robotics, Artificial Intelligence and Real-time Systems at the Technical University of Munich and at the BMW Group. He received his bachelor and master of science in electrical engineering and information technology from the Karlsruhe Institute of Technology, Germany, in 2012 and 2015, respectively. Currently, he is working on interaction-aware behavior prediction of traffic participants and on motion planning under uncertainty in the area of autonomous vehicles. His research interests include robotics, decision making, Bayesian inference and machine learning, with a focus on the intersection of planning and prediction.
Constantin Hubmann, Karlsruhe Institute of Technology and BMW Group
Constantin Hubmann received the bachelor’s and master’s degree in electrical engineering and information technology from Technische Universität München, Munich, Germany, in 2012 and 2014, respectively. He is currently working toward the Doctoral degree with the Chair of Measurement and Control Engineering, Karlsruher Institut für Technologie, Karlsruhe, Germany and in cooperation with the BMW Group. His research interests include maneuver decisions and motion planning under uncertainty for autonomous driving.
Autonomous vehicles have to share the road and interact with various kinds of traffic participants in a social environment. The behavioral uncertainties of other road users make the probabilistic prediction and decision-making extremely challenging. Moreover, modeling human behavior under social interactions is inevitable to enable accurate predictions as well as human-like decision-making and behavioral planning.
In this workshop, we will thoroughly discuss state-of-the-art approaches for probabilistic and socially interactive prediction as well as decision-making and behavioral planning in highly interactive driving scenarios. Also, the fundamental aspects of prediction and decision-making for autonomous driving will be emphasized, such as problem formulation, dataset construction, as well as evaluation and verification, etc.
13:30-13:35, Wei Zhan, Welcome and Overview
13:35-14:10, Prof. Mykel Kochenderfer, Robust Decision Making for Automated Vehicles
14:10-14:35, Constantin Hubmann, Belief State Planning for Interactive Maneuvers in Uncertain Environments
14:35-14:50, Wei Zhan, Behavior Planning of Autonomous Cars with Social Perception
14:50-15:00, Maximilian Graf, A Model Based Motion Planning Framework For Automated Vehicles in Structured Environments
15:00-15:15, coffee break
15:15-15:50, Prof. Katherine Driggs-Campbell, Robust, Informative Behavior Predictions for Trustworthy Autonomy
15:50-16:15, Jens Schulz, Interaction-Aware Probabilistic Behavior Prediction in Urban Environments
16:15-16:30, Zirui Li, Transferable Driver Behavior Learning via Distribution Adaption in the Lane Change Scenario
16:30-16:45, Jiachen Li, Generative Models for Probabilistic Trajectory Prediction
16:45-17:00, Yeping Hu, Interpretable Interactive Behavior Prediction for Autonomous Vehicles
17:00-17:10, Wei Zhan, INTERACTION Dataset: An INTERnational, Adversarial and Cooperative moTION Dataset with Highly Interactive Driving Scenarios
17:10-17:30, panel session
List of Topics
The topics include, but are not limited to:
Social behavior and interactions, modeling and quantification of social factors;
Probabilistic and reactive behavior modeling and prediction for various kinds of traffic participants (vehicles with human drivers, pedestrians, etc.);
Driving data in highly interactive scenarios, collection, processing, annotation and augmentation;
Decision-making under uncertainty, such as POMDP;
Learning algorithms for decision-making and behavior modeling and their applications, such as reinforcement learning, inverse reinforcement learning (IRL), imitation learning;
Deep generative models such as variational auto-encoder (VAE), generative adversarial network (GAN) and their applications;
Bayesian neural network (BNN) and its applications;
Probabilistic graphical models (PGM) for behavior modeling and prediction, such as dynamic Bayesian networks (DBN), hidden Markov models (HMM), etc.;
Combination of learning algorithms and model-based planning/control methods;
Uncertainty modeling and safety verification in deep learning, and applications in behavior modeling and decision-making;
Evaluation, verification/test and failure analysis of prediction or decision-making/planning algorithms in the context of safety-critical systems.