Unverified50% confidenceCautionExact time
time_horizon cannot be shortened indefinitely; for tasks such as quadruped walking that require multi-step coordinated actions, too short a horizon makes it difficult for the agent to observe a complete gait cycle, and the advantage estimate will be biased due to frequent bootstrapping.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe phenomenon where minor deviations at each step of a multi-step task executed by an agent may amplify along the chain and lead to results far from the initial intent is called goal drift.71% similarUnverifiedPure open-loop path planning inherently fails in high-precision assembly tasks, making closed-loop strategies such as force control and visual servoing indispensable in high-precision assembly.70% similarUnverifiedAfter agents are successfully constrained to be on time, they still cannot utilize the extra time to improve task performance, often filling remaining time with repetitive actions.70% similarUnverifiedThe multi-step chained working mode of edge agents causes single-inference latency to be amplified across multiple steps.69% similarUnverifiedA single execution-type agent exhibits three typical problems during long-running tasks: goal drift, shortcut-taking, and mid-task cessation67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1015938API
curl https://kongchang.com/api/v1/knowledge/claims/1015938MCP
get_claim(id=1015938)