2021

HOME: Heatmap Output for future Motion Estimation

Gilles, Thomas, Sabatini, Stefano, Tsishkou, Dzmitry et al.

Understand

In this paper, we propose HOME, a framework tackling the motion forecasting problem with an image output representing the probability distribution of the agent's future location.

This method allows for a simple architecture with classic convolution networks coupled with attention mechanism for agent interactions, and outputs an unconstrained 2D top-view representation of the agent's possible future.
Based on this output, we design two methods to sample a finite set of agent's future locations.
These methods allow us to control the optimization trade-off between miss rate and final displacement error for multiple modalities without having to retrain any part of the model.

Built on

“Layer normalization”
Original
Jimmy Ba, Jamie Kiros and Geoffrey Hinton · 2016
Earlier work this paper cites.
“Desire: Distant future prediction in dynamic scenes with interacting agents”
Namhoon Lee et al · 2017
Earlier work this paper cites.
“Attention is all you need”
A Vaswani et al · 2017
Earlier work this paper cites.
“An intriguing failing of convolutional neural networks and the CoordConv solution”
Rosanne Liu et al · 2018
Earlier work this paper cites.
“R2p2: A reparameterized pushforward policy for diverse, precise generative path forecasting”
Nicholas Rhinehart, Kris Kitani and Paul Vernaza · 2018
Earlier work this paper cites.
“Argoverse: 3d tracking and forecasting with rich maps”
Ming-Fang Chang et al · 2019
Earlier work this paper cites.
“Multimodal trajectory predictions for autonomous driving using deep convolutional networks”
Henggang Cui et al · 2019
Earlier work this paper cites.
“Precog: Prediction conditioned on goals in visual multi-agent settings”
Nicholas Rhinehart, Rowan McAllister, Kris Kitani and Sergey Levine · 2019
Earlier work this paper cites.
“Multiple Futures Prediction”
Yichuan Tang and Ruslan Salakhutdinov · 2019
Earlier work this paper cites.
“Objects as points”
Original
Xingyi Zhou, Dequan Wang and Philipp Krähenbühl · 2019
Earlier work this paper cites.
“nuScenes: A multimodal dataset for autonomous driving”
Holger Caesar et al · 2020
Earlier work this paper cites.
“MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction”
Yuning Chai, Benjamin Sapp, Mayank Bansal and Dragomir Anguelov · 2020
Earlier work this paper cites.

Similar

“Trajectory forecasts in unknown environments conditioned on grid-based plans”
Original
Nachiket Deo and Mohan Trivedi · 2020
Cited alongside, same era.
“Vectornet: Encoding hd maps and agent dynamics from vectorized representation”
Jiyang Gao et al · 2020
Cited alongside, same era.
“Discrete residual flow for probabilistic pedestrian behavior prediction”
Ajay Jain et al · 2020
Cited alongside, same era.
“What-If Motion Prediction for Autonomous Driving”
Original
Siddhesh Khandelwal et al · 2020
Cited alongside, same era.
“The garden of forking paths: Towards multi-future trajectory prediction”
Junwei Liang et al · 2020
Cited alongside, same era.
“Learning lane graph representations for motion forecasting”
Ming Liang et al · 2020
Cited alongside, same era.
“Probabilistic Multi-modal Trajectory Prediction with Lane Attention for Autonomous Vehicles”
Original
Chenxu Luo, Lin Sun, Dariush Dabiri and Alan Yuille · 2020
Cited alongside, same era.
“From Goals, Waypoints & Paths To Long Term Human Trajectory Forecasting”
Original
Karttikeya Mangalam, Yang An, Harshayu Girase and Jitendra Malik · 2020
Cited alongside, same era.
“It is not the journey but the destination: Endpoint conditioned trajectory prediction”
Karttikeya Mangalam et al · 2020
Cited alongside, same era.
“Multi-head attention for multi-modal joint vehicle motion forecasting”
Jean Mercat et al · 2020
Cited alongside, same era.
“Multi-Head Attention with Joint Agent-Map Representation for Trajectory Prediction in Autonomous Driving”
Original
Kaouther Messaoud, Nachiket Deo, Mohan Trivedi and Fawzi Nashashibi · 2020
Cited alongside, same era.
“6th Place Solution: Very Custom GRU”
Ahmet Erdem
Cited in the paper.

Then

“Deep learning-based vehicle behavior prediction for autonomous driving applications: A review”
Sajjad Mozaffari et al · 2020
Later among the works it cites.
“Covernet: Multimodal behavior prediction using trajectory sets”
Tung Phan-Minh et al · 2020
Later among the works it cites.
“Scene compliant trajectory forecast with agent-centric spatio-temporal grids”
Daniela Ridel, Nachiket Deo, Denis Wolf and Mohan Trivedi · 2020
Later among the works it cites.
“Perceive, predict, and plan: Safe motion planning through interpretable semantic representations”
Abbas Sadat et al · 2020
Later among the works it cites.
“Map-Adaptive Goal-Based Trajectory Prediction”
Lingyao Zhang et al · 2020
Later among the works it cites.
“TNT: Target-driven trajectory prediction”
Hang Zhao et al · 2020
Later among the works it cites.
“Argoverse motion forecasting competition” Accessed: 2021-03-12, https://eval.ai/web/challenges/challenge-page/454/leaderboard/1279
2021
Closest in time.
“Multimodal Motion Prediction with Stacked Transformers”
Original
Yicheng Liu et al · 2021
Closest in time.
“Learning to Predict Vehicle Trajectories with Model-based Planning”
Original
Haoran Song et al · 2021
Closest in time.
“TPCN: Temporal Point Cloud Networks for Motion Forecasting”
Original
Maosheng Ye, Tongyi Cao and Qifeng Chen · 2021
Closest in time.
“LaneRCNN: Distributed Representations for Graph-Centric Motion Forecasting”
Original
Wenyuan Zeng, Ming Liang, Renjie Liao and Raquel Urtasun · 2021
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…