DSDNet: Deep Structured self-Driving Network

2020/08/13 by Wenyuan Zeng, Zeng, Wenyuan, Shenlong Wang +9 · 10 citations
Computer Science · Engineering · #Anomaly Detection Techniques and Applications #Autonomous Vehicle Technology and Safety #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Video Surveillance and Tracking Methods #cs.CV

paper · pdf · doi:10.48550/arxiv.2008.06041

ECCV 2020

arxiv created 2020/08/13 · openalex publication_date 2020/08/13 · arxiv updated 2020/08/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

In this paper, we propose the Deep Structured self-Driving Network (DSDNet), which performs object detection, motion prediction, and motion planning with a single neural network. Towards this goal, we develop a deep structured energy based model which considers the interactions between actors and produces socially consistent multimodal future predictions. Furthermore, DSDNet explicitly exploits the predicted future distributions of actors to plan a safe maneuver by using a structured planning cost. Our sample-based formulation allows us to overcome the difficulty in probabilistic inference of continuous random variables. Experiments on a number of large-scale self driving datasets demonstrate that our model significantly outperforms the state-of-the-art.

Citations

• Loopy Belief Propagation for Approximate Inference: An Empirical Study
• Maximum Entropy Deep Inverse Reinforcement Learning
• Fully Connected Deep Structured Networks
• End to End Learning for Self-Driving Cars
• Driving Policy Transfer via Modularity and Abstraction
• Deep Structured Prediction with Nonlinear Output Transformations
• IntentNet: Learning to Predict Intention from Raw Sensor Data
• PIXOR: Real-time 3D Object Detection from Point Clouds
• Class-balanced Grouping and Sampling for Point Cloud 3D Object Detection
• Discrete Residual Flow for Probabilistic Pedestrian Behavior Prediction
• V2VNet: Vehicle-to-Vehicle Communication for Joint Perception and Prediction

Cited by

• Perfect Prediction or Plenty of Proposals? What Matters Most in Planning for Autonomous Driving
• SafeCoop: Unravelling Full Stack Safety in Agentic Collaborative Driving
• You Share Beliefs, I Adapt: Progressive Heterogeneous Collaborative Perception
• CoST: Efficient Collaborative Perception From Unified Spatiotemporal Perspective
• Non-differentiable Reward Optimization for Diffusion-based Autonomous Motion Planning
• Interaction-Merged Motion Planning: Effectively Leveraging Diverse Motion Datasets for Robust Planning
• Automated Vehicles Should be Connected with Natural Language
• AirV2X: Unified Air-Ground Vehicle-to-Everything Collaboration
• Toward Fully Autonomous Driving: AI, Challenges, Opportunities, and Needs
• EP-Diffuser: An Efficient Diffusion Model for Traffic Scene Generation and Prediction via Polynomial Representations

Related

• PredictionNet: Real-Time Joint Probabilistic Traffic Prediction for Planning, Control, and Simulation
• Perceive, Predict, and Plan: Safe Motion Planning Through Interpretable Semantic Representations
• Discrete Residual Flow for Probabilistic Pedestrian Behavior Prediction
• DiGNet: Learning Scalable Self-Driving Policies for Generic Traffic Scenarios with Graph Neural Networks
• SSL-Lanes: Self-Supervised Learning for Motion Forecasting in Autonomous Driving