On May 24th, 2018, Samsung Catalyst Fund co-hosted the Mobility Technology Conference, with Jefferies, Maniv Mobility VC and additional partners. This was the first time the event was held in Israel. 350 professionals from around the globe met to discuss disruptive technology in the mobility sector. Representatives of automotive OEMs and Tier-1 suppliers, as well as other industry executives examined business models, investments and M&A, and the roles China and Israel play in this changing landscape.
With recent news about autonomous driving focusing on the difficulty of detecting objects and people, the role of “perception systems” is taking center stage.
Israel is a powerhouse in sensing and perception, with close to 100 companies in the field. Dede Goldschmidt, VP and Managing Director at Samsung Catalyst Fund moderated a conference panel discussing perception technology, with CEOs and co-founders of four world-leading Israeli companies in the space:
- Oren Rosenzweig, Co-Founder and Chief Business Officer at Innoviz Technologies, the provider of a commercially available solid state LiDAR. Innoviz was recently awarded the LiDAR business for BMW’s Level 3/4/5 program with SOP 2021.
- Hagai Zyss, CEO at Autotalks, a leading provider of vehicle-to-everything (V2X) connectivity technology for safety-critical messaging.
- Igal Raichelgauz, Co-Founder and CEO at Cortica, a company developing autonomous AI technology for unsupervised, highly scalable self-learning.
- Eran Shir, Co-Founder and CEO at Nexar, a company developing a safe driving network of road users. Its community has already driven more than 100 million miles, adding 10 million miles a month.
The panel debated trends in perception technologies. Panelists agreed that current state of AI is not sufficiently advanced to let cars drive around by themselves. There’s a need for adding prediction capabilities to anticipate behaviors of other road users, such as human drivers and pedestrians. This calls for AI technologies that will work the way our brain processes information and creates representation of the world, and that will feature real-time learning in order to manage complex situation on the road. Another key challenge is cumbersome training, which still requires a high degree of supervision, including the heavy lifting of data collection and annotation; and inference accuracy which still does not meet the required quality. The panel concluded that there is definitely room for innovation in this space.
Autonomous cars have the potential to be significantly safer than human drivers. The panel agreed that a wide range of sensors may be needed, complemented by sensor fusion and AI, providing redundant means of determining the car’s surroundings – a camera may become less effective in certain scenarios such as rain, snow, or darkness, but LiDAR and other technologies will back them up and augment, creating a more reliable system. Additional sensors beyond camera will help the autonomous car to not only see things, but also understand them.
The network effect – sharing information between road users – also promises to increase road safety by anticipating the behavior of surrounding actors, especially in non-line-of-site situations. For example, with vehicle-to-vehicle (V2V) technology implemented, your car could know if another car is going to be coming out of an alley just as you arrive. Eventually, it seems likely that all road users will share information, resulting in a safer, more accurate road experience.
Finally, since commercialization of fully automated cars will take years to realize, the panel discussed the introduction of subsets of technologies – primarily around perception, for advanced driver assistance systems (ADAS) – targeting Level 2+ functionality. The key question raised was balancing of features that are truly safety-related, as opposed to gimmicks that might lull the human driver into prematurely relying on Level 2 car automation.
A related keynote at the event was given by Georg Kopets, Co-founder and Member of the Executive Board of TTTech. TTTech is an Austrian company disrupting car infrastructure by building a functional safety software communication fabric for autonomous cars. Georg presented TTTech’s technology, which has already been adopted by Audi for the first commercially-available Level 3-capable car, the Audi A8.
Georg presented the strategic partnership of TTTech with Samsung’s autonomous driving program, Samsung DRVLINE, following a € 75 million investment by Samsung in TTTech. This investment, announced in 2017, was led by the SSIC Smart Machines team in San Jose, California.
We were very happy to co-host this conference, and hope all the participants came away edified by the discussions and with a better feel for the state of automated mobility.