bdd.berkeley.eduBerkeley DeepDrive | We seek to merge deep learning with automotive perception and bring computer vi

") Jump to navigation Enter your keywordsAbout Projects Researchers Sponsors Guiding the Next Generation of Research We are at the forefront of research on deep automotive perception through the integration of two very important technologies: vision and vehicles. Our Mission We seek to merge deep learning with automotive perception and bring computer vision technology to the forefront. ABOUT DEEPDRIVE We’re driving the future of automotive perception. The BDD Industry Consortium investigates state-of-the-art technologies in computer vision and machine learning for automotive applications. Our multi-disciplinary center is housed at the University of California, Berkeley and is directed by Professor Trevor Darrell, Faculty Director of PATH, Professor Kurt Keutzer and Dr. Ching-Yao Chan. The BDD consortium partners with private industry sponsors and brings faculty and researchers together from multiple departments and centers to develop new and emerging technologies with real-world applications in the automotive industry. Research is proposed by UC Berkeley faculty and approved by a BDD advisory board made up of faculty and sponsor representatives. Although dramatic progress has been made in the field of computer visioning, many of these technologies and theories have yet to carry over to the automotive field. Thus, the need and driving force behind the Berkeley DeepDrive Center. Featured projects We are excited about our work. See why. Deep Reinforcement Learning By watching many videos of moving objects, the team’s new tracker learns the relationship between appearance and motion that allows it to track new objects at test time. This tracker thus establishes a new framework for tracking in which the relationship between appearance and motion is learned offline in a generic manner. Cross-modal Transfer Learning This project seeks to transfer models for vision tasks like object detection, segmentation, fine-grained categorization and pose-estimation trained using large-scale annotated RGB datasets to new modalities with no or very few such task-specific labels. Clockwork FCNs for Fast Video Processing The team will use a novel, frame-asynchronous, clockwork” convnet, which pipelines processing over time with different network update rates at different levels of the semantic processing hierarchy, significantly reducing frame processing time with little or no degradation in performance. ALL projects See all of the ways we’re driving the future. Deep Reinforcement Learning Cross-modal Transfer Learning Understanding Driver Awareness for Smart Vehicles Unsupervised Representation Learning for Autonomous Driving Learning to Drive Under Unstructured Conditions Improving the Scaling of Deep Learning Networks by Characterizing and Exploiting Soft Convexity An Innovative Approach to the Dual Problems of High-Resolution Input and Video Object Detection Advisable Deep Driving Hybrid Confidence – Based Human Driver Modeling Explainable Deep Vehicle Control and User-Defined Constraints Learning Human-Like Decision-making Behavior based on Adversarial Inverse Reinforcement Learning Superhuman Vision for Superhuman Driving Hybrid Human Driver Predictions Intelligent Intersection Risk-Averse, Adversarial Reinforcement Learning Embedded natural language processing for in-car speech commands Model-Based Reinforcement Learning Approach for Trajectory and Intent Prediction of Vulnerable Road Users Common Representations for Perception, Prediction, and Planning Hierarchical Reinforcement Learning & Abstraction Discovery A Programming Language for Differentiable Image Processing Multi-modal Fusion of Deep Convolutional Neural Networks for 3D Object Detection Trust-Region Based Robustness of Neural Networks in the Face of Adversarial Attacks Learning to Compile Mobile and Embedded Vision Code with Halide Deep RL with Dexterous Hands and Tactile Sensing AutoPylot: An Open Platform for Autonomous Vehicles Self-supervised Representation Learning for Autonomous Driving Autonomous Driving in Unstructured Stochastic Intersections The Dynamical View of Machine Learning Systems Learning Urban Driving Policies from Online Traffic Webcams Automated Search for Neural Net Architectures Modeling and Learning Multi-Agent Behaviors for Simulation and Analysis of Autonomous Driving Scenarios Fundamental Tradeoffs in Learning and Control Impacts on Energy Consumption of Autonomous Vehicles in the Bay Area Automated Search for Neural Net Architectures Adversarially Robust Visual Understanding Operationalized Active Learning Implementing and Evaluating Second-Order Optimization for Deep Learning on Diverse Architectures Safe and Effective Learning and Control through Formal Simulation Learning Dynamic Point Set Neighbourhoods for 3D Object Detection Investigating driver’s attention while monitoring an autonomous vehicle Maneuver Control based on Reinforcement Learning for Automated Vehicles in an Interactive Environment Interaction and Communication Between Pedestrians and Autonomous Vehicles Autonomous Aerial Robots in Dense Urban Environments Perception-Driven Safe Autonomy in Uncertain Environments Computing Confidence in Human Driver and Pedestrian Models Uncertainty-Aware End-to-End Prediction for Robust Decision Making Life-Long Learning to Drive by Semi-Supervised Reinforcement Learning Human Action and Intention Prediction in Urban Environments Model-Free Reinforcement Learning Through the Optimization Lens Self-supervision, Meta-supervision, Curiosity: The Harder to Learn, the Easier to Generalize Deep Learning-Based Vehicle Control Strategist for Autonomous Vehicles Towards ​safe​ self-driving by reinforcement learning with maximization of diversity of future options Model-based Reinforcement Learning in a Latent Space Fusion of Deep Convolutional Neural Networks for Semantic Segmentation and Object Detection Localization and 3D Detection via Fusing Sparse Point Cloud and Information from Visual-Inertial Navigation System Towards Safe, but not Overly Conservative, Autonomous Cars Bringing in the Human Factor to Resolve Three Fundamental Problems of Deep Learning in Autonomous Driving Guaranteed Adaptation and Exploration in Uncertain Environments Data Augmentation via Synthetic Point Cloud for 3D Detection Refinement and Domain Adaptation with Different LiDAR Configurations Pixel-Level Confidence Prediction for Interpretable Network-Based Driving (ConfPix) Robust Visual Understanding in Adversarial Environments Graph-to-Graph Transfer in Geometric Deep Learning Data-Driven Synthesis of Hazardous Scenarios at Traffic Intersections Robust Perception for Autonomous Driving Model-Based Reinforcement Learning Which Intersections Need I2V and When? Combining Deep Learning and Model Predictive Control for Low-Cost ECUs Self-driving by multi-objective reinforcement learning with goal-conditioned policies Combining Deep Learning and Model Predictive Control for Safe, Effective Autonomous Driving Efficient Neural Networks through Systematic Quantization Enabling Valuable Training Data through Domain Adaptation Explainable Deep Vehicle Control Offline vs. Online Learning for Deep Driving from Demonstrations Spatial and Visual Memory for Navigation Imitative Models: Learning Flexible Driving Models from Human Data Completed Projects See all of the ways we’re driving the future. Clockwork FCNs for Fast Video Processing Deep Reinforcement Learning based Optimization of Autonomous Vehicle Traffic Verifiable Control for (Semi)Autonomous Cars that Learns from Human (Re)Actions Secure and Privacy-Preserving Deep Learning Predictable and Customizable Autonomous Driving Augmenting Autonomous Vehicle Technology with I2V Information FPGA PRET Accelerators of Deep Learning Classifiers for Autonomous Vehicles Fast Simultaneous Object Detection and Segmentation Real-Time Perception/ Prediction of Traffic Scene Outdoor Semantic Scene Segmentation via Multi-modal Sensor Motion Prediction for...