Industry Forum
Partnering for modular, open solutions to meet industry needs
Thursday, May 19, 08.50 — 16.00, SWCC Room 35/36
Organizer: Combient
Agenda
For questions regarding the agenda please contact Stefan Farestam (stefan.farestam@combient.com).
8.50 – 10.10 Session 1
From Closed to Open Research
Stefan Grufman, Team Leader Software & Signal, Husqvarna Group
Husqvarna has been developing different products for more than 300 years, and actually robotic lawn mowers for more than 20 years. Traditionally all R&D has been executed in-house at the Huskvarna site. The climate is changing towards more open collaborations and research. This has lead to a refinement in the Husqvarna research strategy when it comes to robotics. In this talk we will share some of our efforts/thoughts regarding Open Research.
Modular Approach Towards Success
François Boucher, VP Business Development, Kinova Robotics
Kinova will highlight how product and business modularity significantly improves performances in all activities of a company. From product development speed to customer support efficiency, client s and partners can benefit from a modular approach towards success.
PAL Robotics Modular Platforms for Industry 4.0
Carlos Vivas, Business Manager, PAL Robotics
PAL Robotics’ wide experience in humanoid biped robots is used by the company to develop other robotic platforms with specific applications. TIAGo is one of them: a mobile manipulator with applications in industry, that comes with a modular architecture and open-source software to perform any task needed. PAL Robotics’ perspective on the future industry needs will be presented, as well as TIAGo and other robotic platforms developed by the company.
Computer Vision: Enabling Intelligent Consumer Robotics
Dr. Alexander Kleiner, Senior Principle Scientist, iRobot
Worldwide more than 14 million iRobot Roomba floor vacuuming robots have been sold so far. In this talk I give an overview on the currently integrated and future computer vision/AI technologies for the latest model Roomba 980. SLAM has been studied in academia for decades and has recently found its way into commercial consumer products. The Roomba 980 builds a map of its surroundings. It uses this map for systematic cleaning and collision free navigation. Future challenges include long-term mapping and context understanding in households.
10.30 – 11.50 Panel
Partnering for Modular, Open Solutions to Meet Industry Needs
Panelists:
Curtis Carson, Head of Research & Technology – Industrial Strategy & Systems, Airbus
Alisa Adel, Partnerships and Business Development, Airbus
Cristina Cristalli, Research for Innovation Director, Loccioni Group
Kevin Albert, CEO, Pneubotics
Michael Woogerd, Owner, MTM Robotics
Moderator:
Sabine Hauert – University of Bristol & Robohub
With a backlog of nearly 7000 aircrafts on order, Airbus is looking for innovative solutions to increase productivity and improve the quality of life of workers. Drilling robots such as the ones being designed for the Airbus shopfloor challenge could vastly help on both fronts. To have an impact, these robots will need to be quick to deploy and adapt, safe, and precise. All features provided by modular, lightweight, and open platforms. This vision is part of an overall push towards Industry 4.0 where workers and robots operate side by side in a fulfilling and effective manner. There are many challenges ahead however. In this panel discussion, we motivate the need for automation, demystify the current state of the art in manufacturing, and highlight the many technical challenges ahead. We also analyse results from the AIRBUS shopfloor challenge, and extract key lessons on how to translate the technology, while taking into account regulatory and societal barriers.
13.30 – 14.50 Session 3
NVIDIA embedded platform with Deep Learning for next-generation intelligent, autonomous machines
Pierre-Antoine Beaudoin, Embedded PR and Ecosystem EMEA, NVIDIA
NVIDIA® has pioneered visual computing –the art and science of computer graphics– for more than 20 years Today the NVIDIA Jetson TX1 embedded platform, supported by NVIDIA’s complete development tool chain, brings supercomputing power to an embedded platform, driving breakthrough advances in deep learning, artificial intelligence, autonomous machines, robotics and more.
Teaming up With Researchers to Meet Industry Needs
Stefan Lampa, CEO, KUKA Roboter GmbH
KUKA is one of the world’s leading suppliers of robotics and automation solutions. To maintain technological leadership KUKA is closely monitoring market trends and teaming up with researchers around the world to bring new ideas to the market. These innovations are the driving force for the growth of our company. In this talk we will present key success factors for collaborating with researchers, achieving innovations, strengthening our product portfolio and opening up new markets.
Sub-millimeter Accurate Mapping and Tracking for Robots, Drones and AR/VR
Dr. Anna Petrovskaya, Founder and Chief Science Officer, Eonite Perception Inc.
Our goal is to be everyone’s perception module. To this end, we developed an advanced mapping and tracking system with sub-millimeter accuracy. The system is based on commodity depth sensors and runs in real time with very low latency. The super high accuracy opens up a number of new possibilities, with applications in industrial and service robotics, drones, AR/VR, medical robotics, and many others.
DJI Visual Navigation Technologies and Research Platform M100
Shut Yang, Director of Intelligent Navigation Technology, DJI
DJI creates Phantom 4, world first consumer product with mature visual navigation technologies. Its fundamental technologies are developed on Matrice 100, a research oriented drone platform, and onboard SDK.
15.10 – 16.00 Session 4
Deep Learning 2.0: Redesign neurons to minimize training data demand
Bragi Lovetrue, Founder & CTO, Demiurge Technologies AG
The shortage of quality training data is a pressing challenge faced by the robotics and AI industry. It will remain as a long-term bottleneck for robotic and AI applications because large-scale generation, collection, labeling of training data is labor-intensive, time-consuming and cost-inefficient. The root cause of the data shortage challenge, however, is not low data supply but high data demand. A redesign of the neuronal units of deep neural networks could reduce the data demand to a fraction, and Demiurge is at the forefront of developing new neuronal models for Deep Learning 2.0 based on cutting-edge neuroscience.
Socially Intelligent Robots, the next generation of Consumer Robots
Amit Kumar Pandey, Head Principal Scientist (Chief Scientist) & Scientific Coordinator – Collaborative Projects, Aldebaran Robotics, SoftBank Group, Paris, France
The talk will reinforce that the humanoid robots have a range of potential societal applications, and that as a robotics industry, Aldebaran’s R&D and Innovation is around the centrality of wellbeing of people. The first part of the talk will illustrate some of the use cases Aldebaran is seeing for humanoid robot companion. The second part will highlight some of the research directions and results towards achieving such social humanoid robot companion. And last part will present the feedback from real users and conclude by pointing some ethical issues and the grand challenges ahead.