Deep kinematic inference affords efficient and scalable control of bodily movements

Proceedings of the National Academy of Sciences, Journal Year: 2023, Volume and Issue: 120(51)

Published: Dec. 12, 2023

Performing goal-directed movements requires mapping goals from extrinsic (workspace-relative) to intrinsic (body-relative) coordinates and then motor signals. Mainstream approaches based on optimal control realize the mappings by minimizing cost functions, which is computationally demanding. Instead, active inference uses generative models produce sensory predictions, allows a cheaper inversion However, devising complex kinematic chains like human body challenging. We introduce an architecture that affords simple but effective via easily scales up drive chains. Rich can be specified in both using attractive or repulsive forces. The proposed model reproduces sophisticated bodily paves way for efficient biologically plausible of actuated systems.

Language: Английский

---

Embodied decisions as active inference

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: June 1, 2024

Abstract Decision-making is often conceptualized as a serial process, during which sensory evidence accumulated for the choice alternatives until certain threshold reached, at point decision made and an action executed. This decide-then-act perspective has successfully explained various facets of perceptual economic decisions in laboratory, dynamics are usually irrelevant to choice. However, living organisms face another class – called embodied that require selecting between potential courses actions be executed timely dynamic environment, e.g., lion, deciding gazelle chase how fast do so. Studies reveal two aspects goal-directed behavior stark contrast view. First, processes can unfold parallel; second, action-related components, such motor costs associated with required “change mind” them, exert feedback effect on taken. Here, we show these signatures emerge naturally active inference framework simultaneously optimizes perception action, according same (free energy minimization) imperative. We optimizing choices requires continuous loop planning (where beliefs about guide dynamics) finesse alternatives). Furthermore, our simulations normative character ecological settings namely, achieving effective balance high accuracy low risk losing valid opportunities.

Language: Английский

---

Dynamic planning in hierarchical active inference

Neural Networks, Journal Year: 2025, Volume and Issue: 185, P. 107075 - 107075

Published: Jan. 8, 2025

By dynamic planning, we refer to the ability of human brain infer and impose motor trajectories related cognitive decisions. A recent paradigm, active inference, brings fundamental insights into adaptation biological organisms, constantly striving minimize prediction errors restrict themselves life-compatible states. Over past years, many studies have shown how animal behaviors could be explained in terms inference - either as discrete decision-making or continuous control inspiring innovative solutions robotics artificial intelligence. Still, literature lacks a comprehensive outlook on effectively planning realistic actions changing environments. Setting ourselves goal modeling complex tasks such tool use, delve topic keeping mind two crucial aspects behavior: capacity understand exploit affordances for object manipulation, learn hierarchical interactions between self environment, including other agents. We start from simple unit gradually describe more advanced structures, comparing recently proposed design choices providing basic examples. This study distances itself traditional views centered neural networks reinforcement learning, points toward yet unexplored direction inference: hybrid representations models.

Language: Английский

---

Stereo Vision–Based Icicle Detection for Tunnel Deicing

Transportation Research Record Journal of the Transportation Research Board, Journal Year: 2025, Volume and Issue: unknown

Published: April 24, 2025

Icicle damage is common in tunnels cold regions, posing a major threat to the operational safety of railway trains. To address many drawbaks existing deicing methods, this paper proposes design tunnel vehicle with an icicle detection and positioning system. Firstly, structure designed for scenario which operates. Moreover, experiments are conducted based on You Only Look Once (YOLO) deep-learning method. In addition, ice carried out stereo vision, relative error within 2.2%, proving feasibility The combination binocular vision system enables detection, localization, cleaning icicles. This process reduces manual intervention, improves efficiency quality operations, ensures operations.

Language: Английский

---

Modeling Motor Control in Continuous Time Active Inference: A Survey

IEEE Transactions on Cognitive and Developmental Systems, Journal Year: 2023, Volume and Issue: 16(2), P. 485 - 500

Published: Dec. 4, 2023

The way the brain selects and controls actions is still widely debated. Mainstream approaches based on Optimal Control focus stimulus-response mappings that optimize cost functions. Ideomotor theory cybernetics propose a different perspective: they suggest are selected controlled by activating action effects continuously matching internal predictions with sensations. Active Inference offers modern formulation of these ideas, in terms inferential mechanisms prediction-error-based control, which can be linked to neural living organisms. This article provides technical illustration models continuous time brief survey solve four kinds control problems; namely, goal-directed reaching movements, active sensing, resolution multisensory conflict during movement integration decision-making motor control. Crucially, Inference, all facets emerge from same optimization process - minimization Free Energy do not require designing separate Therefore, unitary perspective various aspects inform both study biological design artificial robotic systems.

Language: Английский

---

Learning and embodied decisions in active inference

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 21, 2024

Abstract Biological organisms constantly face the necessity to act timely in dynamic environments and balance choice accuracy against risk of missing valid opportunities. As formalized by embodied decision models, this might require brain architectures wherein decision-making motor control interact reciprocally, stark contrast traditional models that view them as serial processes. Previous studies have assessed dynamics emerge naturally under active inference – a computational paradigm considers action perception subject same imperative free energy minimization. In particular, agents can infer their targets using own movements (and not only external sensations) evidence, i.e., via self-evidencing . Such shown appropriate conditions, action-generated feedback stabilize improve However, how adaptation internal environmental contingencies influences decisions is yet be addressed. To shed light on challenge, study we systematically investigate learning an model during two-alternative forced task, hybrid (discrete continuous) framework. Our results show adapt contexts various statistical regularities task namely, prior preferences for correct target, cue validity, response strategies prioritize faster or slower (but more accurate) decisions. Crucially, these illustrate efficacy discrete sensorimotor from continuous dynamics.

Language: Английский

---

Pose Estimation of a Cobot Implemented on a Small AI-Powered Computing System and a Stereo Camera for Precision Evaluation

Biomimetics, Journal Year: 2024, Volume and Issue: 9(10), P. 610 - 610

Published: Oct. 9, 2024

The precision of robotic manipulators in the industrial or medical field is very important, especially when it comes to repetitive exhaustive tasks. Geometric deformations are most common this field. For reason, new vision techniques have been proposed, including 3D methods that made possible determine geometric distances between parts a manipulator. aim work measure angular position arm with six degrees freedom. purpose, stereo camera and convolutional neural network algorithm used reduce degradation caused by errors. This method not intended replace encoders, but enhance accuracy compensating for through an intelligent visual measurement system. tested about one millimeter. implementation leads better results than traditional simple methods.

Language: Английский

---

Learning and Embodied Decisions in Active Inference

Communications in computer and information science, Journal Year: 2024, Volume and Issue: unknown, P. 72 - 87

Published: Dec. 30, 2024

Language: Английский

---