Autonomous Driving Research

Overview

Autonomous Driving has become one of the most anticipated technologies in both industry and academic research groups. Most of the current efforts in autonomous driving have found success in idealistic conditions such as sparse and homogeneous traffic on highways and urban areas. The GAMMA group, instead, aims to advance autonomous driving research in highly dense and heterogeneous traffic conditions that characterizes social and psychological aspects of human drivers in uncertain environments.

Datasets

  • TRAF Dataset: We provide a dataset of dense and heterogeneous traffic videos. The dataset consists of the following road-agent categories – car, bus, truck, rickshaw, pedestrian, scooter, motorcycle, and other roadagents such as carts and animals. Overall, the dataset contains approximately 13 motorized vehicles, 5 pedestrians and 2 bicycles per frame, respectively. Annotations were performed following a strict protocol and each annotated video file consists of spatial coordinates in pixels, an agent ID, and an agent type. The dataset is categorized according to camera viewpoint (front-facing/top-view), motion (moving/static), time of day (day/evening/night), and difficulty level. The dataset consists of RGB videos with 720p resolution.

  • METEOR: We provide a dataset of dense and heterogeneous traffic videos. The dataset consists of the following road-agent categories – car, bus, truck, rickshaw, pedestrian, scooter, motorcycle, and other roadagents such as carts and animals. Overall, the dataset contains approximately 13 motorized vehicles, 5 pedestrians and 2 bicycles per frame, respectively. Annotations were performed following a strict protocol and each annotated video file consists of spatial coordinates in pixels, an agent ID, and an agent type. The dataset is categorized according to camera viewpoint (front-facing/top-view), motion (moving/static), time of day (day/evening/night), and difficulty level. The dataset consists of RGB videos with 720p resolution.

Software

  • TrackNPred: TrackNPred consists of implementations of state-of-the-art tracking and trajectory prediction methods and tools to benchmark and evaluate them on real-world dense traffic datasets.

  • Behavior-Driven Traffic Simulator: A traffic simulation software based on SUMO that incorporates driver behavior.

Publications

Project Conference/Journal Year
Task-Driven Domain-Agnostic Learning with Information Bottleneck for Autonomous Steering ICRA 2024
Auxiliary Modality Learning with Generalized Curriculum Distillation ICML 2023
METEOR: A Massive Dense & Heterogeneous Behavior Dataset for Autonomous Driving ICRA 2023
Human Trajectory Prediction via Neural Social Physics ECCV 2022
M3DeTR: Multi-representation, Multi-scale, Mutual-relation 3D Object Detection with Transformers WACV 2022
GameOpt: Optimal Real-time Multi-Agent Planning and Control for Dynamic Intersections ITSC’22 2022
GANav: Group-wise Attention Network for Classifying Navigable Regions in Unstructured Outdoor Environments RAL/IROS 2022
Inverse Reinforcement Learning with Hybrid-weight Trust-region Optimization and Curriculum Learning for Autonomous Maneuvering IROS 2022
Gradient-Free Adversarial Training Against Image Corruption for Learning-based Steering NIPS 2021
B-GAP: Behavior-Rich Simulation and Navigation for Autonomous Driving RA-L 2020
Improving Generalization of Transfer Learning Across Domains Using Latent Features in Autonomous Driving Under Review 2021
SS-SFDA : Self-Supervised Source-Free Domain Adaptation for Road Segmentation in Hazardous Environments ICCV 2021
BoMuDA: Boundless Multi-Source Domain Adaptive Segmentation in Unconstrained Environments ICCV 2021
Enhanced Transfer Learning for Autonomous Driving with Systematic Accident Simulation IROS 2020
CMetric: A Driving Behavior Measure using Centrality Functions IROS 2020
Forecasting Trajectory and Behavior of Road-Agents Using Spectral Clustering in Graph-LSTMs RAL/IROS 2020
GraphRQI: Classifying Driver Behaviors Using Graph Spectrums. ICRA 2020
RoadTrack: Tracking Road-Agents in Dense and Heterogeneous Traffic ICRA 2020
RobustTP: End-to-End Trajectory Prediction for Heterogeneous Road-Agents in Dense Traffic with Noisy Sensor Inputs. ACM CSCS 2019
AADS: Augmented Autonomous Driving Simulation Using Data-Driven Algorithms Science Robotics 2019
DensePeds: Pedestrian Tracking in Dense Crowds Using Front-RVO and Sparse Features IROS 2019
TraPHic: Predicting trajectories of Road-Agents in Dense and Heterogeneous Traffic CVPR 2019
ADAPS: Autonomous Driving Via Principled Simulations ICRA 2019
TrafficPredict: Trajectory Prediction for Heterogeneous Agents AAAI 2019
PORCA: Modeling and Planning for Autonomous Driving among Many Pedestrians IEEE Robotics & Automation Letters 2018
AutoRVO: Reciprocal Collision Avoidance between Heterogeneous Agents and Applications to Autonomous Driving AAMAS, CSCS 2018
Identifying Driver Behaviors using Trajectory Features for Vehicle Navigation IROS 2018
Socially Invisible Robot: Navigation in the Social World using Robot Entitativity IROS 2018
Socially Invisible Navigation for Intelligent Vehicles IROS 2018
Classifying Driver Behaviors for Autonomous Vehicle Navigation IROS 2018
AutonoVi: Autonomous vehicle planning with dynamic maneuvers and traffic constraints IROS, CVPR 2017, 2018