Organizations: Department of Electrical and Computer Engineering, National University of Singapore, Singapore · Department of Mechanical Engineering, National University of Singapore, Singapore · Department of Computer Science, National University of Singapore, Singapore · Department of Electrical and Computer Engineering, National University of Singapore · Department of Electrical Engineering, City University of Hong Kong, Hong Kong SAR, China · Advanced Robotic Centre, National University of Singapore, Singapore
Reach-and-attach of soft robotic arms with passive suction requires accurate targeting and sufficient contact speed, yet geared actuators can impose a speed limit that improved trajectory tracking alone cannot overcome. This paper proposes an embodied snap controller that separates slow servo-driven preloading from rapid elastic release, enabling a compliant arm to move beyond its direct tendon-driven speed limit. Octopus biology motivates the controller's section-wise organizational prior, rather than reproduction of the octopus nervous system. A learned policy shared across three sections selects preloads, aim, tendon slack, and release timing, determining where, how, and when to load and release the body. The policy is optimized offline using a hardware-validated recurrent model within experimentally supported bounds. Across five optimization seeds and 400 unseen simulated targets, attachment success is (73±4)% at a 5 cm lateral tolerance, and the shared policy reaches the matched centralized controller's mean final reward after a median 17% of the common evaluation budget. Hardware characterization achieves tip speeds of 1.56-1.64 m/s, at least 108% above direct tendon-driven release. In 18 open-loop hardware trials across six placements, 17 exceed the 1 m/s snap threshold and nine retrieve the object, with successful retrieval at five placements. These results demonstrate a practical division of responsibility in the control problem: learned control prepares the body, and passive body mechanics execute the rapid movement needed for dynamic reach-and-attach.
Figures & tables
Fig. 1: Embodied snap-control system. (a) Three-section soft arm. (b) Simulated primitive: preload toward target, then release S3, S2, and S1 sequentially. (c) A target-conditioned shared policy assigns section-specific preload, aim, slack, and release commands for a feedforward snap and passive attachment.
Fig. 2: Actuation constraint and attachment evidence. (a) Tendon-driven angular-rate ceilings from calibrated tendon kinematics and the rated 12-V no-load servo speed. (b) Tip-speed distributions for ramped and instantaneous distal release. The shaded interval is the direct-release range and the dashed line is the empirical 1m/s attachment threshold. (c) Outcomes from 90 target trials over tip speed and cup-axis tilt. Contours show the fitted two-variable model.
Fig. 3: Hardware snap preconditions and timing. (a) Estimated probability of exceeding 1m/s over proximal preloads; the star is the chosen operating point. (b) Peak tip speed over S1 release delay. (c) Closest approach and attachment outcome in the same sweep. Shaded regions mark the empirically supported operating intervals.
Fig. 4: Hardware gates for controller development. (a) Validation-set open-loop tip error for persistence and the GRU. (b) Measured and predicted peak speed over the excluded delay sweep. (c) Measured and predicted tip travel in the same sweep, bands in (b,c) span model seeds.
Evidence
Domain
n
Use
Separation
Servo calibration
HW
291
Actuator limit
Independent recordings
Attachment trials
HW
90
Speed/angle condition
Run-grouped cross-validation
Delay sweep
HW
33
Timing/dynamics gate
Excluded from model fitting
Dynamics corpus
HW
538
Hold and snap models
Group split: 440 train, 78 validation, 20 sealed test
Policy targets
Sim.
400
Generalisation
Disjoint after 64-target selection
Release targets
Sim.
400
Timing ablation
Offsets fitted on 100 other targets
TABLE I: Evidence sets, their roles, and separation between development and evaluation.
Fig. 5: Architecture benchmark under a common simulation protocol. (a) Held-out reward versus evaluations; lines are means and bands span optimisation seeds. (b) Simulated attachment rate on 400 unseen targets versus geometric tolerance (mean ± standard deviation across seeds). (c) Evaluations required to reach the matched centralised policy’s mean final reward; crosses mark seeds that do not reach it within budget.
Fig. 6: Release-rule ablation in simulation. (a) S1 timing reward; error bars are target-family bootstrap 95% intervals. (b) Simulated attachment rate at the 5cm reporting tolerance ( n=400 ); bars are Wilson 95% intervals and annotations show the fraction projected to a validated timing bound. (c) Simulated attachment rate versus geometric tolerance for the learned policy and fixed primitives.
Fig. 7: Representative successful execution of the frozen learned policy in hardware (top) and simulation (bottom): hanging arm, preload, elastic snap, target reach, and passive attachment.
Evidence
Criterion
Result
Preload reached
S2 bend increase ≥110\text{,}\mathrm{\SIUnitSymbolDegree}$$
18/18
Fast snap produced
Peak tip speed ≥1\text{,}\mathrm{m}\mathrm{/}\mathrm{s}$$
17/18
Snap-speed distribution
Median [IQR]
1.33m/s [1.19–1.44]
Object retrieved
Detached and carried
9/18
Spatial replication
Placements with ≥1 retrieval
5/6
Online correction
State feedback during snap
None
TABLE II: Hardware evidence chain for the learnt policy.
This paper proposes SoftGM, an octopus-inspired distributed control architecture for segmented soft robotic arms that learn to reach targets in contact-rich environments using online obstacle discovery without assuming full obstacle geometry at the start of each episode. SoftGM formulates each arm section as a cooperative agent and represents the arm-environment interaction as a graph. SoftGM uses a two-stage graph attention message passing scheme following a Centralised Training Decentralised Execution (CTDE) paradigm with a centralised critic and decentralised actor. We evaluate SoftGM in a Cosserat-rod simulator (PyElastica) across three tasks that increase the complexity of the environment: obstacle-free, structured obstacles, and a wall-with-hole scenario. Compared with six widely used MARL baselines (IDDPG, IPPO, ISAC, MADDPG, MAPPO, MASAC) under identical information content and training conditions, SoftGM matches strong CTDE methods in simpler settings and achieves the best performance in the wall-with-hole task. Robustness tests with observation noise, single-section actuation failure, and transient disturbances show that SoftGM maintains comparable performance under non-ideal simulated conditions while keeping control effort bounded, suggesting that selective contact-relevant information routing improves resilience in the tested settings.
Linxin Hou, Qirui Wu, Zhihang Qin +2
Department of Electrical and Computer Engineering, National University of Singapore, Singapore · Department of Computer Science, National University of Singapore, Singapore · Department of Mechanical Engineering, National University of Singapore, Singapore +2
High-performance closed-loop control of truly soft continuum manipulators has remained elusive. Experimental demonstrations have largely relied on sufficiently stiff, piecewise architectures in which each actuated segment behaves as a distributed yet effectively rigid element, while deformation modes beyond simple bending are suppressed. This strategy simplifies modeling and control, but sidesteps the intrinsic complexity of a fully compliant body and makes the system behave as a serial kinematic chain, much like a conventional articulated robot. An implicit conclusion has consequently emerged within the community: distributed softness and dynamic precision are incompatible. Here we show this trade-off is not fundamental. We present a highly compliant, fully continuum robotic arm - without hardware discretization or stiffness-based mode suppression - that achieves fast, precise task-space convergence under dynamic conditions. The platform integrates direct-drive actuation, a tendon routing scheme enabling coupled bending and twisting, and a structured nonlinear control architecture grounded in reduced-order strain modeling of underactuated systems. Modeling, actuation, and control are co-designed to preserve essential mechanical complexity while enabling high-bandwidth loop closure. Experiments demonstrate accurate, repeatable execution of dynamic Cartesian tasks, including fast positioning and interaction. The proposed system achieves the fastest reported task-execution speed among soft robots. At millimetric precision, execution speed increases nearly fourfold compared with prior approaches, while operating on a fully compliant continuum body. These results show that distributed compliance and high-performance dynamic control can coexist, opening a path toward truly soft manipulators approaching the operational capabilities of rigid robots without sacrificing morphological richness.
Vito Daniele Perfetta, Daniel Feliu-Talegon, Ebrahim Shahabi +1
Cognitive Robotics Department, Faculty of Mechanical Engineering, Delft University of Technology, Delft 2628 CD, The Netherlands.
Soft continuum robots can exploit distributed compliance for whole-body manipulation, but synthesizing behavior through changing contacts remains difficult. We address whole-body grasping and pick-and-throw from an initially ungrasped state through outcome-based actuation-space optimization. Grasping is quantified by tip angular sweep and body-object enclosure, while throwing further incorporates release-direction alignment and minimum release speed. These objectives allow grasping, acceleration, and release to emerge from compliant interaction without prescribing contact forces, contact locations, or body configurations. Because the resulting actuation-to-outcome mapping is nonsmooth, we utilize derivative-free CMA-ES within an optimize-learn-refine framework. CMA-ES generates solutions for sampled conditions, a task-conditioned predictor learns warm starts, and CMA-ES refines them for unseen conditions. In simulation, the method achieves 492/500 successful grasps (98.4%) and success rates of 98%, 97%, and 94% across three directional throwing trials. Learned initialization increases grasping success from 78.6% to 98.4% while reducing the median rollout count from 1184 to 816 in CMA-ES. Hardware experiments achieve a 100% grasping success rate across 50 executions and a 100% pick-and-throw success rate across 30 executions, with 10 repetitions per direction. Together, these simulation and hardware results demonstrate the effectiveness of the proposed framework across both simulated and physical whole-body manipulation tasks.