arXiv:2605.24950v1 Announce Type: cross Abstract: We present ARCANE-PedSynth, an open-source CARLA-based software framework for generating synthetic multi-pedestrian datasets with dense behavioural annotations for pedestrian crossing prediction in autonomous driving. The framework overcomes CARLA's native 9% crossing rate through a hybrid AI-manual pedestrian control architecture, enabling configurable target rates up to 75%. A 12-state behavioural finite state machine with five character archetypes produces diverse crossing behaviours. The framework generates synchronised RGB, LiDAR, and DVS
The increasing complexity of autonomous driving systems requires more robust and diverse training data, and synthetic data generation tools like ARCANE-PedSynth are emerging to meet this need.
Improving the accuracy of pedestrian prediction in autonomous vehicles is critical for safety and broad adoption, directly impacting the development timeline and reliability of autonomous driving technology.
The ability to generate high-quality, behaviorally diverse synthetic pedestrian data at scale significantly reduces reliance on costly and time-consuming real-world data collection for this specific domain.
- · Autonomous vehicle developers
- · AI model trainers
- · Simulation software providers
- · Pedestrian safety advocates
- · Companies heavily reliant on real-world pedestrian data collection
- · Developers with less sophisticated simulation capabilities
Improved pedestrian prediction in autonomous vehicles leads to safer operations and faster deployment.
Reduced accident rates involving pedestrians could increase public trust and accelerate regulatory approval for autonomous driving.
The methodology could be extended to other complex real-world scenarios, accelerating AI development across various robotic and perception tasks.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG