During the past year our research fellows from the 71 department of the Russian Artificial Intelligence Research Institute along with their collegues from the Moscow Institute of Physics and Technology invested a lot of time in automatic path planning, a very important area in modern applied science which covers self-driving cars, mobile robots, drones and more. Their success in developing methods and algorithms for solving such non-trival problems is now echoed in the publications in the leading scientific journals and proceedings of the international conferences.
As such, the article Policy Optimization to Learn Adaptive Motion Primitives in Path Planning with Dynamic Obstacles (link) by Brian Angulo, Aleksandr Panov and Konstantin Yakovlev was published in IEEE Robotics and Automation Letters, one of the world's leading journals in robotics and automation. The authors introduce a novel learnable steering function (Policy Optimization that Learns Adaptive Motion Primitives) which takes into account kinodynamic constraints of the robot and both static and dynamic obstacles (in other words, how a car-like robot operates in the obstacle-rich parking-lot environments).
The work TransPath: Learning Heuristics for Grid-Based Pathfinding via Transformers (link) by Daniil Kirilenko, Anton Andreychuk, Aleksandr Panov and Konstantin Yakovlev is dedicated to a similar matter (follow the link to watch Daniil's presentation). It was noted by the Association for the Advancement of Artificial Intelligence, chosen from 8777 applicants to be exhibited during the Bridge Program of the annual Conference on Artificial Intelligence AAAI 2023, one of the top-tier international academic conferences in its area (A* Core rating) which will take place in Washington, D.C. from 7 to 14 February.
Same conference will be also exhibiting another collective effort, Safe Interval Path Planning with Kinodynamic Constraints (link) by Zain Alabedeen Ali and Konstantin Yakovlev. It welcomes a new powerful algorithm for solving single-agent pathfinding problem when the agent is confined to a graph and certain vertices/edges of this graph are blocked at certain time intervals due to dynamic obstacles that populate the environment. You may learn more during February the 12th, in Washington or online.