DataTalks HFA #16: Autonomous Systems - Robotics and Planning

📢 Get ready for a one-of-a-kind meetup by Datahack Haifa! Once again we collaborate with the fantastic community of MATAM Gav-Yam, for an event about autonomous systems and robotics. Prof. Indelman will share about approaches developed in his lab which allow autonomous systems to plan and act under uncertainty. Later, Dr. Dotan Di Castro of Bosch's AI research lab will present a framework that enables robots to process language instructions.

• 16:30 MPB (mingling, pizza, beer)
• 17:00 Vadim Indelman
• 17:30 Dotan Di Castro

♦ Location: Tech&Talk MATAM
♦ Language: The event will be held in Hebrew
♦ Background: Basic knowledge in data science and machine learning is advised
------------------------------------------------------------------------------------------------
Abstracts:

🚀 A Glimpse into Online Planning Under Uncertainty and Autonomous Perception
Decision-making under uncertainty is a fundamental problem in robotics and AI. It is at the heart of any autonomous system acting with imperfect information, with numerous applications such as autonomous driving, information gathering, robotic manipulation, autonomous semantic perception, and active SLAM. This talk will provide an overview of recent single- and multi-agent approaches developed in the Autonomous Navigation and Perception Lab (ANPL) that address this important problem from different perspectives.

Vadim Indelman is currently an Associate Professor in the Department of Aerospace Engineering at the Technion - Israel Institute of Technology, and he is also a member of the Technion Autonomous Systems Program (TASP), the Technion Artificial Intelligence Hub (Tech.AI), and the Israeli Smart Transportation Research Center (ISTRC). Additionally, he is a member of the European Laboratory for Learning and Intelligent Systems (ELLIS). He is currently leading the Robotics vertical at TechAI, which promotes and facilitates research activities and projects within the Technion and collaboration with industry in areas related to AI and robotics. Dr. Indelman is the founder and head of the Autonomous Navigation and Perception Lab (ANPL), which performs research related to single and multi-robot autonomous perception, navigation and planning under uncertainty in the context of mobile robotics. His current research interests include planning under uncertainty, probabilistic inference, semantic perception and simultaneous localization and mapping (SLAM) in single and multi-robot systems.

🤖 AI and robotics: Insertion and Language Driven Assembly
In the rapidly advancing domain of robotics, language-driven methods have transformed the way machines understand and execute human commands. Despite the growth of generalist models capable of a wide range of tasks, the precision and reliability required in the domain of industrial robotics poses distinct challenges. In this talk, we will present HiLDA (Hierarchical Language-Driven Assembly), a new hierarchical framework that is designed to process language instructions and trigger specialized skills such as insertion. Our contributions include a centralized controller for task activation, a grasping model for downstream insertion with high positional uncertainty management, an insertion model using online learning for precision, and a comprehensive dataset of insertion trajectories, promising to push forward language-driven robotics in industrial applications

Dr. Dotan opened the AI lab in Haifa in 2019 and serves today as the site manager at Bosch AI in Haifa. The lab consists of 25 scientists and students. Prior to Bosch Dotan was a senior research scientist at Yahoo! research and worked for various startups. Dotan got his Ph.D from the EE faculty in Technion at 2010. His expertise is in AI in general and reinforcement learning (RL) in particular. He has written 40 papers and 35 patents.