AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

DD-PPO: Learning Near-Perfect PointGoal Navigators from 2.5 Billion Frames

Erik Wijmans, Abhishek Kadian, Ari Morcos, Stefan Lee, Irfan Essa, Devi Parikh, Manolis Savva, Dhruv Batra

ICLR 2020 · Reinforcement learning › Visual and representation RL · approach: RL for combinatorial routing · 35 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add synchronous gradient all-reduce with rollout preemption for stragglersPPO training in resource-intensive Habitat simulation · Homogeneous workload, 256 GPUsspeedup: 196xImproved
preempt straggler rollouts at 80% or 60% completion threshold instead of waiting for all workersDD-PPO · Heterogeneous Gibson and Matterport3D workloadscalingImproved
train on Gibson-4+ plus all Matterport3D scenes instead of Gibson-4+ alone (ablation)DD-PPO ResNet50 RGB-D · Habitat Challenge 2019 valSPL: 0.922 ± 0.004 -> 0.956 ± 0.002Improved
add lower-quality Gibson-2/3 reconstructions to training (ablation)DD-PPO ResNet50 RGB-D · Habitat Challenge 2019 valSPL: 0.922 ± 0.004 -> 0.956 ± 0.003Improved
replace ResNet50 with SE-ResNeXt50 (ablation)DD-PPO RGB-D trained on Gibson-2+ · Habitat Challenge 2019 valSPL: 0.956 ± 0.003 -> 0.959 ± 0.002Improved
increase visual encoder capacity to SE-ResNeXt101 and LSTM hidden size to 1024-d (ablation)DD-PPO RGB-D trained on Gibson-2+ · Habitat Challenge 2019 valSPL: 0.959 ± 0.002 -> 0.969 ± 0.002Improved
train RGB navigation with Gibson-2+, all Matterport3D, and channel-wise RGB normalizationDD-PPO RGB agent · Habitat Challenge 2019 valSPL: 0.57 -> 0.929Improved
remove GPS+Compass and train the RGB agent for 2.5 billion stepsDD-PPO navigation · PointGoal navigation without GPS+CompassSPL: 0 -> 0.15Improved
initialize the visual encoder from PointGoalNav rather than ImageNet pre-trainingFlee and Exploration transfer agents · Flee and Explorationtask performanceImproved
initialize and fine-tune the PointGoalNav policy for a new taskFlee and Exploration transfer agents · Flee and Explorationlearning speedImproved

Later papers that compared with this method

3 later accepted papers report DD-PPO as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy