One Thousand and One Hours: Self-driving Motion Prediction Dataset

Abstract

Motivated by the impact of large-scale datasets on ML systems we present the largest self-driving dataset for motion prediction to date, containing over 1,000 hours of data. This was collected by a fleet of 20 autonomous vehicles along a fixed route in Palo Alto, California, over a four-month period. It consists of 170,000 scenes, where each scene is 25 seconds long and captures the perception output of the self-driving system, which encodes the precise positions and motions of nearby vehicles, cyclists, and pedestrians over time. On top of this, the dataset contains a high-definition semantic map with 15,242 labelled elements and a high-definition aerial view over the area. We show that using a dataset of this size dramatically improves performance for key self-driving problems. Combined with the provided software kit, this collection forms the largest and most detailed dataset to date for the development of self-driving machine learning tasks, such as motion forecasting, motion planning and simulation. The full dataset is available at http://level5.lyft.com/.

arXiv commentPresente at CoRL2020

References in corpus (7)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• The ApolloScape Open Dataset for Autonomous Driving and its Application
• Scalability in Perception for Autonomous Driving: Waymo Open Dataset
• Argoverse: 3D Tracking and Forecasting with Rich Maps
• MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction
• ChauffeurNet: Learning to Drive by Imitating the Best and Synthesizing the Worst
• Spatially-Aware Graph Neural Networks for Relational Behavior Forecasting from Sensor Data

Cited by in corpus (25)

• Milestones in Autonomous Driving and Intelligent Vehicles: Survey of Surveys
• Explanations in Autonomous Driving: A Survey
• A Survey on Approximate Edge AI for Energy Efficient Autonomous Driving Services
• FollowNet: A Comprehensive Benchmark for Car-Following Behavior Modeling
• A Unified Longitudinal Trajectory Dataset for Automated Vehicle
• Robots That Can See: Leveraging Human Pose for Trajectory Prediction
• Rethinking Closed-loop Training for Autonomous Driving
• DurLAR: A High-fidelity 128-channel LiDAR Dataset with Panoramic Ambient and Reflectivity Imagery for Multi-modal Autonomous Driving Applications
• Ad-datasets: a meta-collection of data sets for autonomous driving
• NuPlan: A closed-loop ML-based planning benchmark for autonomous vehicles
• VCVW-3D: A Virtual Construction Vehicles and Workers Dataset with 3D Annotations
• Evaluating Object (mis)Detection from a Safety and Reliability Perspective: Discussion and Measures
• MODISSA: a multipurpose platform for the prototypical realization of vehicle-related applications using optical sensors
• Multimodal Trajectory Prediction via Topological Invariance for Navigation at Uncontrolled Intersections
• MS3D++: Ensemble of Experts for Multi-Source Unsupervised Domain Adaptation in 3D Object Detection
• Differentiable Inference of Temporal Logic Formulas
• Reimagining an autonomous vehicle
• Pseudo-labeling for Scalable 3D Object Detection
• Simulation-Based Counterfactual Causal Discovery on Real World Driver Behaviour
• Vehicle Telematics Via Exteroceptive Sensors: A Survey
• Interpretable and Flexible Target-Conditioned Neural Planners For Autonomous Vehicles
• DA-BEV: Unsupervised Domain Adaptation for Bird's Eye View Perception
• Autonomy 2.0: Why is self-driving always 5 years away?
• Scalable Primitives for Generalized Sensor Fusion in Autonomous Vehicles
• DriverGym: Democratising Reinforcement Learning for Autonomous Driving