Autonomous Vehicles Tech Digest - March 2017

Autonomous vehicles edging closer

Volkswagen (VW) has unveiled its autonomous car concept called Sedric. Volkwagen says that this is the first of its cars to be designed for level 5 autonomous driving. Sedric is being envisioned as a mobility solution: Sedric would be called by pressing a button on a dongle-like device. A circular LED display on the dongle informs the user of Sedric’s arrival time with colour-coded signals. Sedric recognises the user and opens its two-part doors to allow entry. When in the car, the user can interact with Sedric by giving a destination, planning stops on the way etc. Once the journey starts the passenger can be entertained through a transparent OLED screen. There is no indication from VW of a general release date. VW displayed a prototype of the chassis and interior of the concept car at the International Motor Show 2017 in Geneva, Switzerland. 
Nvidia, an American graphic processing unit manufacturer, and PACCAR, a truck producer, have produced a proof-of-concept level 4 self-driving truck using Nvidia’s Drive PX 2 technology. The truck has been demonstrated driving on a closed course, and handling various situations without a driver.  
Another company to unveil an autonomous vehicle is Naver, a South Korean ICT company. It has been widely reported that at the Seoul Motor Show 2017 the company showed off a level 3 autonomous car that has already seen testing with a human supervisor on South Korean roads. Naver plans to create a level-4 autonomous car in 2018. 

Intel acquires Mobileye

Intel has announced that it will acquire Mobileye, a company specialising in computer vision for autonomous vehicles. An Intel subsidiary will begin a tender offer to acquire all the shares of Mobileye for USD63.54 per share, which equates to a total equity value of USD15.3 billion and enterprise value of USD14.7 billion. The acquisition will enable Intel and Mobileye to create automated driving solutions encompassing all the key technology components from the cloud through the network to the car, according to Intel’s press release. 

California loosening self-driving laws

Reuters reports that California State will allow testing of autonomous cars without a driver on public roads by the end of 2017. The state is also polling public opinion about scrapping regulations that force the inclusion of manual controls in self-driving vehicles, with a public hearing expected to occur in April 2017.

ARM enters autonomous vehicle market; Bosch and Nvidia work on AI

Microchip manufacturer ARM has unveiled a new chip specifically for autonomous vehicles, the DynamiIQ. ARM claims the processing chip should guarantee a vehicle will react safely in the event of failure even with no human driver to take control, improve performance and efficiency. 
Meanwhile, Bosch and Nvidia have collaborated to release the Bosch AI Car Computer that incorporates Nvidia’s DRIVE PX platform including Xavier technology. Nvidia’s press release states that to achieve a level 4 autonomous car the Xavier technology’s speed (30 trillion deep learning operations a second) and low power consumption (30 watts of power) are needed. Nvidia said it would deliver technology for autonomous vehicle level 4 capabilities by the end of 2018. Bosch’s part in the collaboration would seem to be providing the sensors on autonomous cars for the Nvidia chipset to work with.  

Toyota and NTT to cooperate on 5G connectivity in cars

NTT, a Japanese telecoms operator, and Toyota have announced plans to cooperate on 5G connected car technology research. The research is expected to cover areas such as using 5G for information gathering on traffic conditions and exchanging data between self-driving vehicles to enable safe operations.  

GATEway project continues

In London the GATEway Project (Greenwich Automated Transport Environment) is a project aiming to introduce autonomous vehicles into the Greenwich area. The latest stage of the project is the introduction of a prototype driverless shuttle that will travel a two-kilometre route around the peninsula guided by its Selenium autonomy software. The principal aim of this study is to test how people feel about the vehicle after riding in it. 

Cars seeing in the dark

Researchers at Sookmyung Women’s University and Yonsei University in Seoul have developed a computer vision system to help self-driving cars better see road signs in the dark. The system inputs images captured by the car in real-time and feeds them through a machine learning algorithm. The algorithm judges the reflectiveness of the picture contents, looking for levels of reflectiveness that would correspond to a sign, as well as using the shape to ascertain its message (e.g. a triangular sign will be indicating a warning). Currently the system is able to differentiate irrelevant signs from driving related signs and further divide these into different types, e.g. differentiating 80 miles per hour sign from a 40 miles per hour sign. 

DeepScale

According to multiple sources, Californian start-up DeepScale has raised USD3 million in seed funding to help automotive manufacturers use low-wattage processors to provide better perception. DeepScale has developed a method it says enables deep neural networks and computer vision systems to use less power, making affordable Intel and Qualcomm chips feasible for use in cars. Deepscale said it is working with some OEMs in the automotive industry now and plans to use the same tech in drones and robots in the future. 

UK Government investing in autonomous vehicle avenue

The UK Government has invested GBP100 million in a programme to create an autonomous vehicle business development corridor along the M40 that connects Birmingham and London. The corridor’s purpose is to attract businesses involved in testing autonomous vehicles.  The GBP100 million is intended to be matched by industry to take the total investment to GBP200 million over four years. Businesses can bid for the programme’s investment. 

 

Add this: