Abhijat Biswas

I am a PhD student in the Robotics Institute at Carnegie Mellon University, working with Henny Admoni on using driver eye-gaze information to inform active sensor policies for safer autonomous and assisted driving. I worked with Henny and Aaron Steinfeld, on social navigation and related topics for Human-Robot Interaction during my Masters.

Previously, I've spent time teaching neural nets to recognize free hand sketches at IISc Bangalore. I've also worked on using temporal smoothness in videos as supervision for neural networks at Cardiff University. I went to IIT Guwahati for ECE.

Email  /  CV  /  Google Scholar  /  Github

Ongoing work
  1. Dynamics of peripheral vision during driving
    Right after saccades (eye movements), human peripheral vision degrades. How can driver assistance systems incorporate this knowledge to provide intelligent assistance when driver peripheral awareness is low?
  2. Attention during critical driving situations
    What does eye gaze reveal about driver situational awareness (SA) during safety-critical driving situations? How can we predict when SA does not cover task relevant stimuli and initiate appropriate interventions?
  1. SocNavBench: A Grounded Simulation Testing Framework for Evaluating Social Navigation
    A Biswas, A Wang, G Silvera, A Steinfeld, and H Admoni
    ACM Transactions on Human-Robot Interaction (THRI) 2021
    [Paper (Arxiv)] [Code][Paper (ACM, coming soon)]
  2. Examining the Effects of Anticipatory Robot Assistance on Human Decision Making
    B Newman*, A Biswas*, S Ahuja, S Girdhar, K Kitani, and H Admoni
    International Conference on Social Robotics (ICSR) 2020
  3. Human Torso Pose Forecasting in the Real World
    A Biswas, H Admoni, and A Steinfeld
    Multi-modal Perception and Control Workshop, Robotics:Science and Systems (RSS) 2018
  4. SketchParse: Towards Rich Descriptions for Poorly Drawn Sketches using Multi-Task Hierarchical Deep Networks
    RK Sarvadevabhatla, I Dwivedi, A Biswas, S Manocha, and RV Babu
    ACM Multimedia Conference (ACM MM) 2017
    [Paper] [Code]
  5. Development of an assistive stereo vision system
    T Shankar, A Biswas, and V Arun
    International Convention on Rehabilitation Engineering and Assistive Technology (i-CREATe) 2015
    [Paper] [News]
First-order Meta-Learned Initialization for Faster Adaptation in Deep Reinforcement Learning [PDF]
Abhijat Biswas, Shubham Agrawal
First-derivative approximations to meta-learning updates perform just as well as second-derivative ones. Demonstrated on RL tasks
Socially compliant path planning [Video] [PDF] [Code]
Abhijat Biswas, Ting-Che Lin, Sean Wang
RTAA* + Social-LSTM based social navigation
Automatic Extrinsic Calibration of Stereo Camera and 3D LiDAR [Poster]
Abhijat Biswas, Aashi Manglik
Known RGB-extrinsics can be used to remove human intervention (usually used to get LiDAR-RGB correspondences)