Χωρίς GPS, τα αυτόνομα οχήματα χάνονται εύκολα. Τώρα ένας νέος αλγόριθμος που αναπτύχθηκε στο Caltech (California Institute of Technology) επιτρέπει στα αυτόνομα οχήματα να αναγνωρίζουν πού βρίσκονται απλά κοιτάζοντας το έδαφος γύρω τους – και για πρώτη φορά, η τεχνολογία λειτουργεί ανεξάρτητα από τις εποχιακές αλλαγές σε αυτό το έδαφος.
See also: EU: AI makes autonomous vehicles "extremely vulnerable" to cyberattacks
Details about the process were published on June 23 in the journal Science Robotics, which was published by the American Association for the Advancement of Science (AAAS).
The general process, known as visual terrain-relative navigation (VTRN), was first developed in the 1960s. Autonomous vehicles do not lose their orientation, comparing the nearby terrain with high-resolution satellite images.
See also: Germany: Passes legislation on the circulation of autonomous vehicles
The problem is that, in order to work, the current VTRN generation requires the terrain it examines to match the images in its database. Anything that alters or hides the terrain, such as snow or fallen leaves, causes the images not to match. VTRN systems can operate correctly if the database contains landscape images for every possible weather condition.
To overcome this challenge, Caltech turned to deep learning and artificial intelligence (AI) to remove era-specific content that hinders the current VTRN systems.
The process – which was developed by Soon-Jo Chung and Anthony Fragoso in collaboration with graduate student Connor Lee and undergraduate student Austin McCoy – uses what is known as “self-supervised learning”. While most computer vision strategies rely on people carefully curating large datasets to teach an algorithm how to recognize what it sees, this instead lets the algorithm teach itself. Artificial intelligence searches for patterns in images by pulling details and features that humans are likely to miss.
Beyond its utility for autonomous vehicles on Earth, the system also has applications for space missions. The Entry, Descent, and Landing (EDL) system on Mars 2020 Perseverance rover , for example, used VTRN for the first time on the Red Planet to land in Jezero Crater, a location previously considered too dangerous for a landing. The team looked at areas of Mars that have strong seasonal changes, conditions similar to Earth, and the new system could enable improved navigation to support science objectives, including the search for water.
See also: China: Turns to automation to address population issues
Next, Fragoso, Lee and Chung will expand the technology to also account for weather changes: fog, rain, snow, etc. If successful, their work could contribute to improving navigation systems for driverless cars.
Source of information: caltech.edu
