Latest trends in AI and self-driving revealed at Taiwan+Israel Hi-Tech Forum

By combining Israel's innovation dynamics and Taiwan's IT manufacturing infrastructure, a brand new supply chain and cooperation paradigm could be created in the new era of autonomous driving.

By combining Israel’s innovation dynamics and Taiwan’s IT manufacturing infrastructure, a brand new supply chain and cooperation paradigm could be created in the new era of autonomous driving. To this end, at the “Taiwan+Israel Hi-Tech Forum,” which was jointly organized in late August by the Bureau of Foreign Trade, Ministry of Economic Affairs, and Taiwan External Trade Development Council (TAITRA), three companies from Israel, Mobileye, Check Point and Karamba Security, shared the latest development trends in AI, autonomous driving, and security technologies with an audience from the local tech industries.

James Huang, chairman of Taiwan External Trade Development Council, said in his opening remarks that Israel is a world-renowned nation of startups, while Taiwan has a complete ecosystem covering R&D, design and manufacturing. The combination and complementation of their advantages will open up unlimited possibilities in the latest self-driving sector. In particular, after leading a delegation to visit Israel this May, Huang said he was deeply impressed by its innovation strength, which prompted the arrangement of this forum, which marked an important milestone for the cooperation between Taiwan and Israel in the high-tech industry. TAITRA will also work to enable cooperation and exchanges between the two sides and promote more business opportunities in the future.

Mobileye’s self-driving development strategy

Founded in 1999 and acquired by Intel in 2017, Mobileye is an iconic high-tech company in Israel and a leader in the ADAS (advanced driver assistance systems) market. Rosana Su, general manager of Mobileye (Shanghai), said that ever since the company’s establishment, it has been committed to the development of monocular-based computer vision processing technology. Until now, there are over 27 million Mobileye-equipped vehicles worldwide, and 13 automakers working with Mobileye to enable autonomous driving.

For the development of self-driving, Mobileye’s strategy consists of three major parts: leveraging ADAS technology as building blocks, combining with low-cost camera-centric sensing technology to achieve economic scalability, and formalizing safety to ensure safety.

According to Su, in addition to camera-centric sensing technology, crowd-sourced mapping and semantic driving policy are two important pillars to establish a responsibility-sensitive security model, in a bid to meet consumers’ higher safety demands for autonomous driving.

In terms of roadscape recognition, the advantage of camera sensing technology is that it can provide rich information. However, due to weather conditions or other factors, its field of vision may be affected. Also, localization is required because traffic signs vary from country to country. These are all the challenges that need to be overcome. Therefore, other aided information from LiDAR, ladar, and maps are required in order to fully grasp the traffic conditions.

Among them, HD maps play an important role. Mobileye’s approach is to use its unique, low-cost REM (Road Experience Management) solution to collect landmark information from vehicles equipped with its ADAS, upload it to the cloud, and work with map operators around the world to quickly update the HD map. This crowd-sourcing method, of course, requires accurate collection of extensive data. Currently, Mobileye has partnered with Nissan, Volkswagen and General Motors on this regard.

After combining sensing and map information, the most important part of self-driving is how to develop the right driving policy. Mobileye is currently working with the US Department of Transportation to analyze data of past fatal traffic accidents as a reference and establish guidelines for driving policy.

Statistics show that human error is the cause of most road accidents. By making use of ADAS to provide early warning, 90% of car collisions can be avoided. Mobileye now provides various ADAS solutions for different kinds of fleet. In addition to working with car makers, products for the after-service market are also available.

She stressed that the ultimate goal of developing self-driving cars is to reduce accidents, not to pursue high technology for technology’s sake. Considering technology maturity and cost factors, this will be a gradual evolution process. Therefore, Mobileye will start from ADAS and slowly add more features in order to achieve the L5 self-driving goal in the long run.

Security of connected devices enhanced by innovative technology

Tony Jarvis, chief strategist for Asia Pacific, Middle East and Africa at Check Point Software Technologies, said that AI is progressing rapidly and has been widely adopted by industries. According to statistics, nearly 50% of companies have a defined AI strategy. In fact, AI has undergone long-term development. The recent rapid progress is mainly due to the reduction of storage costs, the improvement of computing power, and the maturity of mathematical algorithms, so that enormous amounts of data can be stored and analyzed.

However, AI is double-edged for cybersecurity. On the one hand, hacking methods keep evolving, with AI-based malware emerging and developing rapidly. On the other hand, proper use of AI technology can provide further security protection.

A good AI security solution requires a large amount of data, combined with AI expertise and domain-specific expertise. But at present the key problems with cybersecurity AI is that both data and expertise are not enough.

Check Point, which specializes in IT security technology, is actively developing AI-related technologies to improve cybersecurity. The company has established three AI security mechanisms, including Campaign Hunting (predictive threat intelligence), Huntress (which uncovers malicious executables), and CADET (context-aware detection). After testing, they can actually improve the protection rate and reduce the false positives.

He stressed that after the era of mobile and cloud, emerging technologies such as blockchain, Internet of Things, and AI will be the mainstream in the future. The importance of cybersecurity technology will become more prominent. Enterprises must grasp this trend and dare to invest in order to maintain their competitive advantages.

Karamba tackles growing risk of cyberattacks

Founded in 2016, Karamba Security, with its unique security technology for connected cars, has raised a second round of US$27 million in just two years and cooperated with leading companies, including Denso, QNX, Honeywell, and Micro.

Amir Einav, VP of Marketing at Karamba, noted that 5G, Wi-Fi, Bluetooth, and DSRC (dedicated short range communications) have become chinks in the armor during security attacks. In addition, with the increasing number of electronic systems in the car, the size of the software code is getting larger and larger, so they are inherently vulnerable. For example, in May of this year, Keen Lab found 14 vulnerabilities in BMW car models, and over the past three years, the ECU (engine control unit) has become a new target for security attacks.

The current architecture of the connected car is a set of IoT devices, and Karamba aims to become the unique cybersecurity enabler in IoT and automotive, in an effort to enhance the security of in-vehicle communication and bring safety to autonomous cars. Its technology includes runtime authentication and runtime integrity. The concept is to protect the memory of the device from any foreign code, so that the device can avoid the attacks of malware by only performing the specific functions that the system defines.

He emphasized that the patent-pending security technology has negligible performance impact to the cars, and it can realize self-protecting SoC/ECU by using new encryption methods and message level encryption technologies. It’s very important for ECU devices with limited resources. Karamba’s “built-in security” concept can effectively prevent attacks, reduce operational costs, and provide automotive-grade solutions.