Both STEP2DYNA and ULTRACEPT are explicitly built around insect visual pathway models — looming and motion-sensitive neurons — applied to collision detection.
AGILE ROBOTS AG
Munich robotics company applying insect-inspired neural vision systems to real-time collision detection for autonomous vehicles and industrial robots.
Their core work
Agile Robots AG is a Munich-based robotics company specializing in intelligent robotic systems with advanced machine perception. Their EU-funded research work centers on translating biological visual processing — specifically how insects detect approaching objects through looming-sensitive neurons — into artificial neural architectures for real-time collision avoidance. They bring industrial robotics engineering into international academic consortia, bridging the gap between theoretical neural computation models and deployable perception systems for robots and autonomous vehicles. Their contributions to H2020 projects focus on multi-modal sensing, including thermal and visual channels, for safety-critical collision detection.
What they specialise in
ULTRACEPT keywords include VLSI, neural system models, and simulation, indicating hands-on work translating neural models into hardware-implementable circuits.
STEP2DYNA is specifically titled around spatial-temporal processing for dynamic environments, forming the theoretical foundation for their applied vehicle safety work.
ULTRACEPT (2018–2024) shifts focus directly to vehicle collision avoidance with multi-modal and thermal perception, targeting autonomous vehicle and ADAS application domains.
ULTRACEPT keywords include thermal image and multiple modality, suggesting capability in sensor fusion beyond standard camera-only systems.
How they've shifted over time
Their early participation (STEP2DYNA, 2016) centered on fundamental research: modelling biological visual systems and realizing spatial-temporal processing architectures for generic collision detection in dynamic environments. By 2018, ULTRACEPT marks a clear pivot toward application — vehicle-specific collision avoidance, multi-modal sensing including thermal imaging, and hardware-level neural models (VLSI). The trajectory runs from abstract bio-inspired modelling toward commercially deployable perception technology for autonomous vehicles and robotic safety systems.
They are moving from foundational neural computation research toward applied, hardware-implementable perception systems for autonomous vehicles — a direction that positions them squarely in the autonomous mobility and industrial robot safety markets.
How they like to work
Agile Robots joins consortia as a participant rather than leading projects, contributing specialized industrial robotics expertise to larger, academically-led research teams. With 19 unique partners from just 2 MSCA-RISE projects, they are comfortable operating in wide, distributed international networks — the staff-exchange format of MSCA-RISE means these relationships involve genuine cross-institutional secondments, not just contractual links. For a consortium builder, they are a focused specialist who will integrate well without requiring coordination overhead.
Across 2 projects, Agile Robots has built connections with 19 unique partners in 6 countries — a broad reach relative to their small H2020 footprint. MSCA-RISE consortia routinely include non-European partners, so their actual network likely spans beyond Europe into Asia or the Americas, though this cannot be confirmed from available data.
What sets them apart
Agile Robots is unusual among robotics companies in holding formal EU research consortium experience specifically in bio-inspired perception — a niche that sits at the intersection of computational neuroscience and industrial robotics engineering. Unlike pure academic groups, they bring product-oriented engineering perspective and deployment know-how to research teams; unlike generic tech companies, they have a demonstrated track record in long-running international consortia (projects running 5–6 years). For anyone building a Horizon Europe consortium around autonomous perception, robot safety, or neuromorphic computing, they offer a credible industry anchor with existing research network depth.
Highlights from their portfolio
- ULTRACEPTTheir largest and most recent project (EUR 279,000, running to 2024), it targets vehicle collision avoidance with multi-modal and thermal perception — the most commercially relevant problem in their portfolio and a direct bridge to autonomous vehicle markets.
- STEP2DYNATheir first H2020 project (2016–2021) established the bio-inspired neural processing foundation that underpins all subsequent work, making it the origin point of their research identity in this domain.