Neural Simultaneous Localization and Mapping Workshop

Simultaneous Localization and Mapping (or SLAM) refers to the problem of constructing a map of an unknown environment as it is actively being explored. SLAM has been treated extensively in mobile robotics, providing a rich theoretical framework within which to understand spatial processing in the brain. Equally, neuroscientific progress has uncovered many of the spatial representations used by the brain such as place, grid, head-direction and boundary cells. The aim of this two-day workshop, held at EITN Paris, will be to survey and contrast topics at the frontiers of SLAM in both robotics and neuroscience to identify novel cross-cutting areas of interest to both domains. Several aspects of function must be shared where both brains and machines are tasked with operating in the same complex environments. Yet, despite their commonalities, theories of navigation in robotics and neuroscience have largely evolved independently. Nonetheless, the rich literature (both experimental and theoretical) on spatial representations in the brain promises to provide novel insights into the SLAM problem. Conversely, robust testing in physically realistic environments and complex tasks, as is typical in the robotics domain and a central vision of the HBP, can further constrain biological models. Talks can consist of theoretical or experimental work if they provide insights into SLAM on a biological level. Approximately 80% of the talks will be invited. Contributions are therefore invited to fill the remaining 20%. Early career researchers are especially encouraged to apply. Topics will include, but are not limited to perspectives on active learning / sensing including reinforcement learning approaches, embodied and neurally-inspired approaches, systems / architectural / anatomical perspectives, front-end processing and multi-sensory integration by machines and brains. Contributions are invited from all sub-domains of the HBP.