VISION-SLS: Safe Perception-Based Control from Learned Visual Representations via System Level Synthesis

Antoine P. Leeman, Shuyu Zhan, Melanie N. Zeilinger, Glen Chou

arXivarxiv/2604.24894Apr 27, 2026cs.ROscore 0

We propose VISION-SLS, a method for nonlinear output-feedback control from high-resolution RGB images which provides robust constraint satisfaction guarantees under calibrated uncertainty bounds despite partial observability, sensor noise, and nonlinear dynamics. To enable scalability while retaining guarantees, we propose: (i) a learned low-dimensional observation map from pretrained visual features with state-dependent error bounds, and (ii) a causal affine time-varying output-feedback policy optimized via System Level Synthesis (SLS). We develop a scalable, novel solver for the resulting…

SourceJSON

Comments and reviews

No comments yet.

API

curl /api/v1/records/arxiv/2604.24894
curl -X POST /api/v1/records/arxiv/2604.24894/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
  -d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2604.24894/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
  -d '{"body": "…", "review": true}'