Back to feed
AR
arXiv CS.AI
7/29/2026
ProcAgent: An Agentic Framework for Procedural Task Guidance on Edge with Human-in-the-Loop

ProcAgent: An Agentic Framework for Procedural Task Guidance on Edge with Human-in-the-Loop

Short summary

ProcAgent is a fully on-device, vision-based agentic assistant for procedural task guidance running on a single NVIDIA Jetson AGX Orin. It uses a propose-and-verify architecture combining continuous perception, a symbolic task graph, on-demand vision-language verification, and an LLM interaction agent. In a 10-participant user study, it achieved positive ratings for comprehensibility, actionability, and privacy comfort while resolving text queries in ~2s and visual queries in ~8s.

  • •Fully on-device agentic procedural assistant running on NVIDIA Jetson AGX Orin
  • •Propose-and-verify architecture with symbolic task graph and on-demand visual reasoning
  • •10-participant user study shows positive ratings; text queries ~2s, visual queries ~8s

Generated with AI, which can make mistakes.

#research-breakthrough#ai-agents#ai-tools#open-source
Read full article at arXiv CS.AI

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

The original title is "AgentGUI: An Interface for Observing and Steering Long-Running AI Agents"
arXiv cs.CL
HyperAgent: Planning and Acting over Tool-Schema Hypergraphs for Tool-Use LLM Agents
arXiv CS.AI
The original title is "Agentless: A Practical, Cost-Effective Workflow for Human-Supervised AI"
Dev.to
Building a VideoAgent-Style Multi-Agent System: Intent Parsing, Graph Planning, and Tool Routing for Video Editing Tasks
MarkTechPost