MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction

Abstract

Predicting human behavior is a difficult and crucial task required for motion planning. It is challenging in large part due to the highly uncertain and multi-modal set of possible outcomes in real-world domains such as autonomous driving. Beyond single MAP trajectory prediction, obtaining an accurate probability distribution of the future is an area of active interest. We present MultiPath, which leverages a fixed set of future state-sequence anchors that correspond to modes of the trajectory distribution. At inference, our model predicts a discrete distribution over the anchors and, for each anchor, regresses offsets from anchor waypoints along with uncertainties, yielding a Gaussian mixture at each time step. Our model is efficient, requiring only one forward inference pass to obtain multi-modal future distributions, and the output is parametric, allowing compact communication and analytical probabilistic queries. We show on several datasets that our model achieves more accurate predictions, and compared to sampling baselines, does so with an order of magnitude fewer trajectories.

Appears in CoRL 2019

References in corpus (7)

• Social Force Model for Pedestrian Dynamics
• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?
• Fast and Furious: Real Time End-to-End 3D Detection, Tracking and Motion Forecasting with a Single Convolutional Net
• CARLA: An Open Urban Driving Simulator
• IntentNet: Learning to Predict Intention from Raw Sensor Data
• ChauffeurNet: Learning to Drive by Imitating the Best and Synthesizing the Worst
• Forecasting Social Navigation in Crowded Complex Scenes

Cited by in corpus (48)

• One Thousand and One Hours: Self-driving Motion Prediction Dataset
• LookOut: Diverse Multi-Future Prediction and Planning for Self-Driving
• Social-WaGDAT: Interaction-aware Trajectory Prediction via Wasserstein Graph Double-Attention Network
• Learning Lane Graph Representations for Motion Forecasting
• What-If Motion Prediction for Autonomous Driving
• Multimodal Motion Prediction with Stacked Transformers
• DenseTNT: End-to-end Trajectory Prediction from Dense Goal Sets
• Maneuver-Aware Pooling for Vehicle Trajectory Prediction
• Heterogeneous Edge-Enhanced Graph Attention Network For Multi-Agent Trajectory Prediction
• PLOP: Probabilistic poLynomial Objects trajectory Planning for autonomous driving
• RetinaTrack: Online Single Stage Joint Detection and Tracking
• V2VNet: Vehicle-to-Vehicle Communication for Joint Perception and Prediction
• DenseTNT: Waymo Open Dataset Motion Prediction Challenge 1st Place Solution
• LiRaNet: End-to-End Trajectory Prediction using Spatio-Temporal Radar Fusion
• Probabilistic Multi-modal Trajectory Prediction with Lane Attention for Autonomous Vehicles
• DSDNet: Deep Structured self-Driving Network
• The Garden of Forking Paths: Towards Multi-Future Trajectory Prediction
• Non-Gaussian Chance-Constrained Trajectory Planning for Autonomous Vehicles under Agent Uncertainty
• Multi-modal Trajectory Prediction for Autonomous Driving with Semantic Map and Dynamic Graph Attention Network
• PRANK: motion Prediction based on RANKing
• TPCN: Temporal Point Cloud Networks for Motion Forecasting
• Improving Movement Predictions of Traffic Actors in Bird's-Eye View Models using GANs and Differentiable Trajectory Rasterization
• Shifts: A Dataset of Real Distributional Shift Across Multiple Large-Scale Tasks
• DiversityGAN: Diversity-Aware Vehicle Motion Prediction via Latent Semantic Sampling
• Multi-modal Scene-compliant User Intention Estimation in Navigation
• SurfelGAN: Synthesizing Realistic Sensor Data for Autonomous Driving
• Spatio-Temporal Graph Dual-Attention Network for Multi-Agent Prediction and Tracking
• Maneuver-based Anchor Trajectory Hypotheses at Roundabouts
• Data Driven Prediction Architecture for Autonomous Driving and its Application on Apollo Platform
• TrafficSim: Learning to Simulate Realistic Multi-Agent Behaviors
• MSN: Multi-Style Network for Trajectory Prediction
• From Recognition to Prediction: Analysis of Human Action and Trajectory Prediction in Video
• AutoTrajectory: Label-free Trajectory Extraction and Prediction from Videos using Dynamic Points
• Efficient Latent Representations using Multiple Tasks for Autonomous Driving
• UST: Unifying Spatio-Temporal Context for Trajectory Prediction in Autonomous Driving
• Ellipse Loss for Scene-Compliant Motion Prediction
• Autonomy 2.0: Why is self-driving always 5 years away?
• Trajectory Prediction with Graph-based Dual-scale Context Fusion
• Prediction-Based Reachability for Collision Avoidance in Autonomous Driving
• Vehicle Trajectory Prediction by Transfer Learning of Semi-Supervised Models
• A Spatial-Temporal Attentive Network with Spatial Continuity for Trajectory Prediction
• Jointly Learning Agent and Lane Information for Multimodal Trajectory Prediction
• Negotiation-Aware Reachability-Based Safety Verification for AutonomousDriving in Interactive Scenarios
• Scenario-Based Trajectory Optimization in Uncertain Dynamic Environments
• Move Beyond Trajectories: Distribution Space Coupling for Crowd Navigation
• Probability Trajectory: One New Movement Description for Trajectory Prediction
• Open-set Intersection Intention Prediction for Autonomous Driving
• Motion Prediction Performance Analysis for Autonomous Driving Systems and the Effects of Tracking Noise