Device-Conditioned Neural Architecture Search for Efficient Robotic Manipulation

arXiv:2604.10170v1 Announce Type: cross Abstract: The growing complexity of visuomotor policies poses significant challenges for deployment with heterogeneous robotic hardware constraints. However, most existing model-efficient approaches for robotic manipulation are device- and model-specific, lack generalizability, and require time-consuming per-device optimization during the adaptation process. In this work, we propose a unified framework named extbf{D}evice-extbf{C}onditioned extbf{Q}uantization-extbf{F}or-extbf{A}ll (DC-QFA) which amortizes deployment effort with the device-conditioned quantization-aware training and hardware-constrained architecture search. Specifically, we introduce a single supernet that spans a rich design space over network architectures and mixed-precision bit-widths. It is optimized with latency- and memory-aware regularization, guided by per-device lookup tables. With this supernet, for each target platform, we can perform a once-for-all lightweight search to select an optimal subnet without any per-device re-optimization, which enables more generalizable deployment across heterogeneous hardware, and substantially reduces deployment time. To improve long-horizon stability under low precision, we further introduce multi-step on-policy distillation to mitigate error accumulation during closed-loop execution. Extensive experiments on three representative policy backbones, such as DiffusionPolicy-T, MDT-V, and OpenVLA-OFT, demonstrate that our DC-QFA achieves 2ext{-}3imes acceleration on edge devices, consumer-grade GPUs, and cloud platforms, with negligible performance drop in task success. Real-world evaluations on an Inovo robot equipped with a force/torque sensor further validates that our low-bit DC-QFA policies maintain stable, contact-rich manipulation even under severe quantization.

Related

• SIM1: Physics-Aligned Simulator as Zero-Shot Data Scaler in Deformable Worlds
• A Physical Agentic Loop for Language-Guided Grasping with Execution-State Monitoring
• LIDEA: Human-to-Robot Imitation Learning via Implicit Feature Distillation and Explicit Geometry Alignment
• Whole-Body Mobile Manipulation using Offline Reinforcement Learning on Sub-optimal Controllers
• StaMo: Unsupervised Learning of Generalizable Robot Motion from Compact State Representation
• EgoVerse: An Egocentric Human Dataset for Robot Learning from Around the World

Source: arXiv cs.CV | 2026-04-14