trajectory transfer
Imagine your friend has solved a tricky puzzle before. Instead of starting from scratch, you borrow their notes on the steps they took, skipping the dead ends. That’s the general idea here: taking a record of a successful sequence of actions, called a trajectory, from one task or agent and using it to improve performance on a related, but different, task.
Across these projects, the winning move isn’t just copying the raw steps. Instead, they find that picking apart what actually caused success, converting steps into reusable summaries or rules, and even replaying them in a new setting beats blindly training on the original log. Sharing these distilled lessons, whether from robots to robots or between digital assistants, cuts wasted effort and boosts final results, even when the source and target agents work differently.
- Papers
- 28
- Released code
- 0
- First seen
- June 2026
- Latest
- Nov 2026
25 papers in the last two months, against 2 in the two before.
Start here
Three papers to read first, in the order they came out- 1Text traces can steer frozen agents without additional training
REVEAL steers frozen open-source agents using textual frontier model trajectories, improving task completion without training or internal access.
Oct 2026 - 2Human demonstrations scale whole-body learning for humanoid loco-manipulation
Robot-aligned whole-body human demonstrations scale humanoid loco-manipulation training, improving task progress and producing strong benchmark and real-world results.
Oct 2026 - 3Training on interfaces helps visual models inspect images beyond screens
GUI grounding supervision trains vision-language models to inspect images sequentially, improving perception beyond GUI tasks.
Oct 2026
How fast it is moving
Often studied with
verificationChecking outputs for correctness before accepting them, often to catch errors or reward hacking.6 togethervision-language-action modelA robot policy model that maps visual observations and language instructions to actions, trained on demonstrations and often used for manipulation tasks.5 togethercredit assignmentDetermining which actions or steps in a sequence contributed to a final outcome, often for reinforcement learning.5 togetherWho is working on it
The papers
Most central to this idea first, not most recent.
- Chinese Techcs.RO
Human demonstrations scale whole-body learning for humanoid loco-manipulation
Robot-aligned whole-body human demonstrations scale humanoid loco-manipulation training, improving task progress and producing strong benchmark and real-world results.
Alibaba Token Hub, Amap · Oct 2026
- Chinese Techcs.AI
Text traces can steer frozen agents without additional training
REVEAL steers frozen open-source agents using textual frontier model trajectories, improving task completion without training or internal access.
Zhejiang University, South China University of Technology · Oct 2026
- Chinese Techcs.CV
Training on interfaces helps visual models inspect images beyond screens
GUI grounding supervision trains vision-language models to inspect images sequentially, improving perception beyond GUI tasks.
Wuhan University of Technology, Institute of Automation, Chinese Academy of Sciences (CAS) · Oct 2026
- Chinese Techcs.LG
Diffusion models learn better short-budget reasoning from their longer-window trajectories
Masked diffusion language models can be improved by distilling their own large-window reasoning into small-window decoding, boosting accuracy and shortening responses.
The Hong Kong University of Science and Technology (Guangzhou), Zhejiang University · Oct 2026
- Independentcs.AI
Rewriting harness-assisted solutions improves general terminal-task performance
Rewriting successful multi-harness agent traces into verified general-harness trajectories substantially improves a base model on complex terminal tasks.
Oct 2026
- Chinese Techcs.SE
Agents learn software repair rules that transfer across repositories
XRepoSkill learns transferable software engineering skills by deriving rules from divergent successful and failed trajectories and verifying them across repositories.
Nanyang Technological University, Tsinghua University · Sept 2026
- Chinese Techcs.RO
Portable RGB demonstrations can train mobile manipulation policies and reduce base-velocity errors
Portable RGB demonstrations via VOMMI train mobile manipulation policies that lower base-velocity error by 18.2% compared to robot-collected demonstrations.
Zhejiang University, Shanghai AI Lab · Oct 2026
- Chinese Techcs.CL
AI agents’ average agreement can mask declines in later browsing actions
TraceSpec shows that allocating draft supervision by trajectory state and prefix-aware acceptance gaps raises agent speculative-decoding throughput by 12.5% to 22.6% over DSpark.
Renmin University of China, Huawei Technologies Co., Ltd. · Nov 2026
- Big Techcs.RO
Robots learn dexterous skills from a single egocentric human video
HandMimicX learns dexterous manipulation from a single human video using object motion and interaction priors for spatial generalization.
Tsinghua University, TsingTec · Nov 2026
- Big Techcs.AI
Governing memory lifecycle, not just retrieval, boosts LLM agent reuse
Lifecycle-controlled memory improves coding-agent retrieval, bug resolution, and cost by structuring experience, gating detailed recall on struggle, and evicting details after use.
Amazon, Yale University · Oct 2026
- Big Techcs.AI
Hypergraphs help language agents reuse feedback across planning paths
HyperMCTS uses a hypergraph to share decision outcomes across trajectories, improving MCTS efficiency and accuracy for LLM agents.
University of Florida, Amazon · Sept 2026
- Top Universitycs.AI
Shared agent memories boost performance without coordination
A shared repository that stores and retrieves agent-generated trajectories improves task performance and reduces interaction steps in ALFWorld and WebArena.
Carnegie Mellon University, University of California, Berkeley · June 2026
- Big Techcs.RO
Contact-preserving retargeting improves robot task success from 13% to 83%
Contact-anchored retargeting with explicit force targets preserves demonstrated contact far better than kinematic methods and improves task success.
Carnegie Mellon University, Amazon Frontier AI & Robotics (FAR) · Oct 2026
- Academiccs.LG
AI agents can drop from 85% to 19% success with rearranged information
A policy trained under one agent harness secretly encodes that harness's action preferences, so changing the interface can destroy performance even when optimal actions stay identical.
The Hong Kong University of Science and Technology (Guangzhou), The Hong Kong University of Science and Technology · Oct 2026
- Academiccs.RO
Multi-agent framework learns to coordinate puzzle piece assembly without planning
Multi-agent reinforcement learning with a graph neural network can coordinate multiple parts to assemble Tangram puzzles simultaneously.
Technical University of Darmstadt · Nov 2026
- Chinese Techcs.LG
Reconstructing app behavior before coding improves simulated GUI agent training
Separating behavior reconstruction from implementation yields faithful GUI environments from traces, with 91.1% workflow fidelity and 94.0% real-device gain retention.
Nanjing University, Alibaba Token Hub, Alibaba Group · Oct 2026
- Big Techcs.CV
AI models can transfer actions between human and robot videos
DreamX repurposes a pretrained video diffusion model to learn compact latent actions that transfer across embodiments and achieve state-of-the-art on LARYBench.
KAIST, NVIDIA · Nov 2026
- Chinese Techcs.LG
Language models can learn from expert guidance when every attempt fails
Policy-Compatible Reinforcement Learning recovers verified training examples for hard problems by guiding a model with expert solutions while limiting how far its behavior shifts.
University of Science and Technology of China, Tencent · Oct 2026
- Big Techcs.SE
Rewarding necessary edits makes debugging patches smaller without reducing correctness
PreciseCoder reduces average edit lines to 3 per bug while keeping pass rates on par with standard training.
University of Chicago, University of Southern California · Oct 2026
- Chinese Techcs.AI
Unified platform enables GUI agent training across mobile and computer environments
MAI-Gym unifies GUI agent evaluation and RL training across mobile and computer environments, achieving state-of-the-art performance with a 9B model.
Alibaba Group · Oct 2026
- Top Universitycs.LG
Specialized GFlowNet mixtures find more modes without extra oracle calls
A population of GFlowNets with a learned classifier divides state space to improve mode coverage beyond single-policy training.
Mila – Québec AI Institute, Intel · Oct 2026
- Chinese Techcs.AI
Intervening During Rollouts Helps Agents Learn Beyond Their Initial Limits
Reward-aware correction of failed rollouts at intervention time lets LLM agents learn from trajectories they could not generate on their own, pushing RL beyond policy limits.
Alibaba Token Hub, Alibaba Group, HKUST (Guangzhou) · Oct 2026
- Top Universitycs.RO
Forecasting 3D scene motion improves dexterous robot manipulation
Jointly forecasting 3D point trajectories of scene and hands improves dexterous robotic manipulation.
POSTECH, KAIST · Oct 2026
- Top Universitycs.RO
Human demonstrations teach humanoids whole-body manipulation without robot teleoperation
Workhorse enables humanoid robots to perform whole-body manipulation tasks by learning from human demonstrations without robot-specific retargeting.
University of California, Berkeley · Oct 2026
- Chinese Techcs.CL
Hybrid agents learn to recreate software across five computing platforms
Training hybrid agents on RecreationWorld trajectories improves performance on out-of-distribution coding and hybrid benchmarks.
Alibaba Token Hub, Alibaba Group · Sept 2026
- Independentcs.AI
Training agents to combine graphical interfaces and commands improves efficiency
CUA-Universe turns real desktop software into hybrid GUI+CLI environments and trains more capable, efficient agents.
Sept 2026
- Industrycs.AI
Switching Models Mid-Task Imposes a Cost and Quality Penalty
Handoffs between cheap and expensive coding agents incur a quality-cost penalty, the handoff tax, which varies with direction and trajectory information.
AWS · Aug 2026
- Independentcs.SE
Existing scientific software can generate verified training tasks for terminal agents
Existing scientific software workflows provide self-supervised training data for terminal agents via software-in-the-loop reconstruction.
Oct 2026
Concepts are extracted from each paper and reused across the corpus, so this page grows on its own as the desk reads.