{"slug": "a-latent-control-model-for-realistic-rodent-motion", "title": "A latent control model for realistic rodent motion", "summary": "A team of neuro-AI researchers posted arXiv paper 2610.27279v1 introducing SCAMPER (Simulated Control via Augmented Motor Priors for Embodied Reinforcement learning), a system that trains an embodied virtual rat model to move naturally both during goal-directed behavior and during random exploration. SCAMPER uses a variational architecture with two distinct priors — a motion prior for control and a behavior prior for exploration — and the authors report that combining both priors lets the system generate realistic exploratory behavior while exploiting task structure, and that the behavior prior's diverse movements enable learning from sparse rewards. The work targets the neuro-AI community's need for embodied animal models whose movements can be directly compared with biological agents.", "body_md": "arXiv:2610.27279v1 Announce Type: new \nAbstract: The development of embodied models of animal behavior is a major goal of the neuro-AI community. Embodied models will enhance our ability to make direct comparisons between biological and artificial agents. To make such comparisons effective, though, we need embodied models that actually move like animals. Previous work has developed virtual animal bodies and techniques for training realistic motion through imitation. However, to date, there is no system for embodied animal modeling that produces natural movements both during trained goal-directed behavior, and during exploration driven by random generation. Here, we introduce Simulated Control via Augmented Motor Priors for Embodied Reinforcement learning (SCAMPER), a system for training an embodied virtual rat model, enabling natural movements during both goal-directed behavior and random exploration. SCAMPER uses a variational architecture with two distinct priors, a motion prior for control and a behavior prior for exploration. We show that combining both of these priors allows SCAMPER to generate realistic behavior during exploration and exploit the task structure effectively. SCAMPER can also learn from sparse rewards as a result of the diverse movements induced by the behavior prior, allowing the system to model realistic sparse-reward tasks from experimental neuroscience. Altogether, SCAMPER provides the neuro-AI community with an effective new system for training embodied models of rodents on realistic tasks.", "url": "https://wpnews.pro/news/a-latent-control-model-for-realistic-rodent-motion", "canonical_source": "https://www.machinebrief.com/news/a-latent-control-model-for-realistic-rodent-motion-4xe9", "published_at": "2026-10-21 04:00:00+00:00", "updated_at": "2026-10-21 04:50:12.386569+00:00", "lang": "en", "topics": ["ai-research", "machine-learning", "robotics", "neural-networks"], "entities": ["SCAMPER", "arXiv", "Simulated Control via Augmented Motor Priors for Embodied Reinforcement learning"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/a-latent-control-model-for-realistic-rodent-motion", "markdown": "https://wpnews.pro/news/a-latent-control-model-for-realistic-rodent-motion.md", "text": "https://wpnews.pro/news/a-latent-control-model-for-realistic-rodent-motion.txt", "jsonld": "https://wpnews.pro/news/a-latent-control-model-for-realistic-rodent-motion.jsonld"}}