A Coordinated Dual-Arm Framework for Delicate Snap-Fit Assemblies

TL;DR

A dynamical-systems-based dual-arm coordination framework is presented that integrates SnapNet driven detection with an event-triggered impedance modulation, enabling accurate alignment and compliant insertion during delicate snap-fit assemblies.

Abstract

Delicate snap-fit assemblies, such as inserting a lens into an eye-wear frame or during electronics assembly, demand timely engagement detection and rapid force attenuation to prevent overshoot-induced component damage or assembly failure. We address these challenges with two key contributions. First, we introduce SnapNet, a lightweight neural network that detects snap-fit engagement from joint-velocity transients in real-time, showing that reliable detection can be achieved using proprioceptive signals without external sensors. Second, we present a dynamical-systems-based dual-arm coordination framework that integrates SnapNet driven detection with an event-triggered impedance modulation, enabling accurate alignment and compliant insertion during delicate snap-fit assemblies. Experiments across diverse geometries on a heterogeneous bimanual platform demonstrate high detection accuracy (over 96% recall) and up to a 30% reduction in peak impact forces compared to standard impedance control.

Figures (9)

• Figure 1: Bimanual snap-fit task performed under a unified framework using Franka FR3 and Addverb Heal Cobot. The proposed method is directly applicable to snap-fit assembly skills for bimanual/humanoid robots.
• Figure 2: Representative perception–planning–control pipeline for automated snap-fit assembly. This study targets the shaded blocks: high-level motion planning, snap-fit engagement detection, and bimanual coordinated insertion. All other modules are treated as given.
• Figure 3: Sensorized Franka Hand used for data collection. In (a), the gripper directly pushes the component into its mating counterpart while in (b), the gripper holds the part and performs the insertion into its mating counterpart.
• Figure 4: Snap signatures noted on the most prominent joints of Franka FR3.
• Figure 5: SnapNet architecture. Per-joint CNN-GRU encoders extract local and temporal features from joint velocities, which are fused via attention and classified through a sigmoid layer.