This project proposes to advance the knowledge and technological development of neuromorphic systems through the use of emerging memristive nanodevices, applying Artificial Intelligence (AI) techniques for traffic sign recognition and obstacle detection in autonomous vehicles. Neuromorphic sensing and processing systems inspired by the human brain have demonstrated great potential for various applications requiring low power consumption and high speed, such as driving control in autonomous vehicles based on visual information generated by multiple sensors. The Artificial Intelligence section of the 2023 Gartner Hype Cycle highlights advances and methodologies that provide substantial advantages and potentially transformative results. It also acknowledges the limitations and vulnerabilities of imperfect systems. According to the Gartner curve, neuromorphic engineering is one of the "hot topics" expected to reach its productivity plateau in the next 5 to 10 years, demonstrating the potential of the paradigm proposed in this project.
The VINDALOO project aims to make significant progress in several areas: 1) developing more efficient vision sensors that incorporate dynamic foveation mechanisms to reduce the amount of data generated outside of areas of interest; 2) developing efficient neuromorphic processors based on hybrid CMOS-memristor technology that use these nanodevices to emulate synaptic connections between neurons; and 3) developing event-based learning algorithms (both supervised and unsupervised) for object detection and tracking applications. The results obtained in these three areas will be experimentally validated using a multi-chip hardware demonstrator that can identify traffic signs and detect obstacles in real time while driving, with low power consumption. The synergy between the various scientific and technological objectives of this project is expected to yield significant advances in the knowledge necessary to translate bio-inspired concepts into competitive final products. The Neuromorphic Systems Group that proposes this project enjoys significant international recognition in this field, with a distinguished track record working on both national and European projects.
Within the framework of Andalusia's Smart Specialisation Strategy for Sustainability 2021-2024, this project falls under Specialisation Environment E1 (Smart, Resilient, and Healthy Society), specifically Subenvironment E1.S3 (Information and Communication Technologies and Digital Content Sector). This subenvironment encompasses advanced technological developments and new paradigms linked to fields such as Artificial Intelligence and nanotechnology, which correspond precisely with the neuromorphic computing systems proposed for development in this project. These systems incorporate memristive nanodevices into a hardware system that uses AI techniques for vision applications. This aligns particularly well with the Challenge of Environment E1.S3.R4, which aims to position Andalusia as a leader in areas such as AI, automation, and robotics.
