LeTS-Drive: Driving in a Crowd by Learning from Tree Search

  title={LeTS-Drive: Driving in a Crowd by Learning from Tree Search},
  author={Panpan Cai and Yuanfu Luo and Aseem Saxena and David Hsu and Wee Sun Lee},
Autonomous driving in a crowded environment, e.g., a busy traffic intersection, is an unsolved challenge for robotics. [] Key Method It consists of two phases. In the offline phase, we learn a policy and the corresponding value function by imitating the belief tree search. In the online phase, the learned policy and value function guide the belief tree search. LeTS-Drive leverages the robustness of planning and the runtime efficiency of learning to enhance the performance of both. Experimental results in…

Figures and Tables from this paper

Closing the Planning-Learning Loop with Application to Autonomous Driving in a Crowd

To achieve real-time performance for large-scale planning, this work introduces a new algorithm Learning from Tree Search for Driving (LeTS-Drive), which integrates planning and learning in a closed loop, and applies it to autonomous driving in crowded urban traffic in simulation.

Monte Carlo Tree Search With Reinforcement Learning for Motion Planning

This work enables real-time applicability of a Monte Carlo Tree Search algorithm with a deep-learning heuristic and benchmark the proposed agent against multiple methods: rules-based, MCTS, $A^{*}$ search, deep learning, and Model Predictive Control.

LEADER: Learning Attention over Driving Behaviors for Planning under Uncertainty

A new algorithm is proposed, LEarning Attention over Driving bEhavioRs (LEADER), that learns to attend to critical human behaviors during planning, using importance sampling to bias reasoning towards critical events.

KB-Tree: Learnable and Continuous Monte-Carlo Tree Search for Autonomous Driving Planning

The results show that the proposed KB-Tree method performs well in all tasks, and outperforms the learning-based continuous MCTS method and the state-of-the-art Reinforcement Learning (RL) baseline.

Socially Aware Crowd Navigation with Multimodal Pedestrian Trajectory Prediction for Autonomous Vehicles

A new method is proposed that combines a deep socially aware attentive value network with a human multimodal trajectory prediction model to help identify the optimal driving policy and introduces a novel technique to extend the discrete action space with minimal additional computational requirements.

Simulating Autonomous Driving in Massive Mixed Urban Traffic

SUMMIT is a high-fidelity simulator that facilitates the development and testing of crowd-driving algorithms and supports a wide range of applications, including perception, vehicle control or planning, and end-to-end learning.

Deep Visual Navigation under Partial Observability

  • Bo AiWei GaoVinayD. Hsu
  • Computer Science
    2022 International Conference on Robotics and Automation (ICRA)
  • 2022
This work proposes to train a neural network (NN) controller for local navigation via imitation learning and integrates the multiple neural network modules into a unified controller that achieves robust performance for visual navigation in complex, partially observable environments.

Decision-making and Planning Framework with Prediction-Guided Strategy Tree Search Algorithm for Uncontrolled Intersections

A cooperative framework composed of a Primary Driver for motion planning and a Subordinate Driver for decision-making, which works as a collision checker and a low-level motion planner to generate a safe and smooth trajectory.

SUMMIT: A Simulator for Urban Driving in Massive Mixed Traffic

This paper presents SUMMIT, a high-fidelity simulator that facilitates the development and testing of crowd-driving algorithms by leveraging the open-source OpenStreetMap map database and a heterogeneous multi-agent motion prediction model developed in earlier work.

Efficient POMDP Behavior Planning for Autonomous Driving in Dense Urban Environments using Multi-Step Occupancy Grid Maps

This paper enables intention-aware POMDP planning while considering uncertainties by using Multi-step Occupancy Grid Maps (MOGM) to represent the current and predicted states of surrounding road users, as well as their uncertain intentions.



Socially aware motion planning with deep reinforcement learning

Using deep reinforcement learning, this work develops a time-efficient navigation policy that respects common social norms and is shown to enable fully autonomous navigation of a robotic vehicle moving at human walking speed in an environment with many pedestrians.

Deep-Learned Collision Avoidance Policy for Distributed Multiagent Navigation

This work presents a novel end-to-end framework to generate reactive collision avoidance policy for efficient distributed multiagent navigation by formulating an agent's navigation strategy as a deep neural network mapping from the observed noisy sensor measurements to the agent's steering commands in terms of movement velocity.

Intention-aware online POMDP planning for autonomous driving in a crowd

This paper presents an intention-aware online planning approach for autonomous driving amid many pedestrians that uses the partially observable Markov decision process (POMDP) for systematic, robust decision making under uncertainty.

DESPOT: Online POMDP Planning with Regularization

This paper presents an online POMDP algorithm that alleviates these difficulties by focusing the search on a set of randomly sampled scenarios, and gives an output-sensitive performance bound for all policies derived from a DESPOT, and shows that R-DESPOT works well if a small optimal policy exists.

HyP-DESPOT: A hybrid parallel algorithm for online planning under uncertainty

HyP-DESPOT is a massively parallel online planning algorithm that integrates CPU and GPU parallelism in a multi-level scheme that speeds up online planning by up to a factor of several hundred in several challenging robotic tasks in simulation, compared with the original DESPOT algorithm.

Thinking Fast and Slow with Deep Learning and Tree Search

This paper presents Expert Iteration (ExIt), a novel reinforcement learning algorithm which decomposes the problem into separate planning and generalisation tasks, and shows that ExIt outperforms REINFORCE for training a neural network to play the board game Hex, and the final tree search agent, trained tabula rasa, defeats MoHex 1.0.

CrowdMove: Autonomous Mapless Navigation in Crowded Scenarios

A generalized yet effective 3M (i.e., multi-robot, multi-scenario, and multi-stage) training framework is proposed and a mapless navigation policy is optimized with a robust policy gradient algorithm.

Planning and Acting in Partially Observable Stochastic Domains

Monte-Carlo Planning in Large POMDPs

POMCP is the first general purpose planner to achieve high performance in such large and unfactored POMDPs as 10 x 10 battleship and partially observable PacMan, with approximately 1018 and 1056 states respectively.

Avoiding cars and pedestrians using velocity obstacles and motion prediction

An iterative planning approach that addresses the problem of estimating the future behaviour of moving obstacles and to use the resulting estimates in trajectory computation and an iterative motion planning technique based on the concept of Velocity Obstacles.