Junhong Xu

Vehicle Autonomy and Intelligence Lab at Indiana University, Bloomington.

profile.jpg

About Me

I am a fifth-year Ph.D. candidate at Vehicle Autonomy and Intelligence Lab at Indiana University, Bloomington. My research focuses on decision-making under uncertainty for real-world robotic applications. I am particularly interested in developing practical and scalable decision-making algorithms (a.k.a. stochastic control – I deliberately omitted “optimal” as it is generally difficult to find an optimal solution for many interesting real-world problems) that consider the following commonly existing uncertainties in real-world robotic systems:

  1. motion outcome uncertainty, where the robot cannot accurately predict its motion outcome after executing a control command,

  2. interaction uncertainty, where the behaviors of other interacting agents are uncertain,

  3. state uncertainty, where the physical state of the robot cannot be accurately determined (e.g., localization error),

  4. perception uncertainty, where the outcome of a perception model is uncertain (e.g., object class cannot be predicted with certainty due to occlusion).

Prior to my Ph.D., I also worked on reducing DAgger’s manual labeling in imitation learning and multi-task imitation learning for mobile robot navigation in indoor environments.

News

Oct, 2022
Our paper "Causal Inference for De-biasing Motion Estimation from Robotic Observational Data" on leveraging causal inference to learn robot motion model is on arxiv (submitted to ICRA 2023).
Sep, 2022
Our paper "Decision-Making Among Bounded Rational Agents" on explicit modeling of computational limits in multi-agent motion planning using information-theoretic bounded rationality is accepted by DARS 2022.
Oct, 2021 Our team is 1 of 4 teams selected (others are MIT, CMU and University of Washington) by the Army Research Laboratory (ARL) to enhance unmanned ground vehicle’s navigation and control capability in off-road environments.
Nov, 2020
We successfully deployed our decision-making under uncertainty planner to off-road environments.
Sep, 2020 Our paper “Online Planning in Uncertain and Dynamic Environment in the Presence of Multiple Mobile Vehicles” is accepted by IROS 2020.
Jul, 2020 Our paper “State-Continuity Approximation of Markov Decision Processes via Finite Element Methods for Autonomous System Planning” is accepted by R-AL.
Jun, 2020
Our paper "Kernel Taylor-Based Value Function Approximation for Continuous-State Markov Decision Processes" on developing scalable algorithm for decision-making under uncertainty problems is accepted in Robotic: Science and Systems.

Selected Publications

  1. Decision-Making Among Bounded Rational Agents
    Junhong Xu, Durgakant Pushp, Kai Yin, and 1 more author
    Distributed Autonomous Robotic Systems 2022
  2. Causal Inference for De-biasing Motion Estimation from Robotic Observational Data
    Junhong Xu, Kai Yin, Jason M Gregory, and 1 more author
    arXiv preprint arXiv:2210.08679 2022
  3. Kernel Taylor-Based Value Function Approximation for Continuous-State Markov Decision Processes
    Junhong Xu, Kai Yin, and Lantao Liu
    Robotics: Science and System 2020
  4. Online Planning in Uncertain and Dynamic Environment in the Presence of Multiple Mobile Vehicles
    Junhong Xu, Kai Yin, and Lantao Liu
    In 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2020
  5. State-continuity approximation of markov decision processes via finite element methods for autonomous system planning
    Junhong Xu, Kai Yin, and Lantao Liu
    IEEE Robotics and Automation Letters 2020
  6. Reachable space characterization of markov decision processes with time variability
    Junhong Xu, Kai Yin, and Lantao Liu
    Robotics: Science and Systems 2019
  7. Automated labeling for robotic autonomous navigation through multi-sensory semi-supervised learning on big data
    Junhong Xu, Shangyue Zhu, Hanqing Guo, and 1 more author
    IEEE Transactions on Big Data 2019
  8. Shared multi-task imitation learning for indoor self-navigation
    Junhong Xu, Qiwei Liu, Hanging Guo, and 3 more authors
    In 2018 IEEE global communications conference (GLOBECOM) 2018
  9. A deep residual convolutional neural network for facial keypoint detection with missing labels
    Shaoen Wu, Junhong Xu, Shangyue Zhu, and 1 more author
    Signal Processing 2018