Autonomous Vehicles Tech Digest - June 2017

Lyft and nuTonomy partner for tests

Lyft, an American ride sharing company, has partnered with nuTonomy, an autonomous car software developer, to carry out field trials of self-driving cars in the Boston, USA, area to gain insight into the passengers’ experience of autonomous cars. The trial will use nuTonomy’s existing fleet of test vehicles in Boston, and build on research from Singapore trials where the company plans to have autonomous ride-sharing services on a commercial basis in 2018. The collaborative trials will test function, performance and features of an on-demand autonomous ridesharing service.

Chinese trambus

CRRC, a Chinese locomotive manufacturer, has revealed a bus-tram hybrid that would run autonomously on specially created lanes in cities called ART (Autonomous Rail Rapid Transit). The modular vehicle can have carriages added or subtracted depending on usage and demand with each carriage having a maximum passenger capacity of 300 people. ART follows lines painted onto the road surface and uses a suite of sensors to avoid obstacles etc. The electric vehicle, that can go 25 km on a single ten-minute charge, is being built in Zhuzhou, China with plans to have it deployed on a 6.5km line in the same city in 2018.

GM claims ability to mass produce autonomous cars

GM claims to have become the first automotive company to have mass produced self-driving vehicles. The statement came as the company announced the mass-manufacture of 130 of its autonomous Chevrolet Bolt EV. GM further claimed that it is the only company that has the ability to produce autonomous cars at scale. The new autonomous Bolts are intended to join GM’s existing fleet of test vehicles in San Francisco, Scottsdale and Michigan.

Audi sets up centre for deep learning

Audi announced that it will work with Austria’s Johannes-Kepler-University Linz (JKU) to develop artificial intelligence (AI) technology to improve autonomous car systems’ self-learning abilities at the Audi JKU Deep Learning Centre. The centre has five initial projects among which are research into autonomous cars’ awareness in complex traffic situations for optimization of hazard awareness and reactivity, and the use of AI in sensors.

Ocado delivery van

The GATEway project, a UK autonomous vehicle project, has trialled autonomous vehicles for grocery delivery from the online retailer Ocado. The trial sees a self-driving delivery van called CargoPod delivering orders to over 100 customers in the Greenwich area of London. The vehicle was created by Oxbotica and uses the company’s Selenium self-driving software platform. It can carry a total of 128kg of groceries at a time.

Dubai Police to buy an autonomous patrol car

The police department of Dubai plans to get an autonomous police patrol car. O-R3 features technologies such as face recognition, live broadcasting, voice call for interaction between citizens and human police. It navigates autonomously using SLAM technology, a front facing laser scanner and thermal camera and a 360-degree LiDAR and HD camera array. The small car comes with a drone that can assist it to carry out surveillance from the air.

5G remote control car

SAIC Motor, a Chinese car manufacturer, China Mobile and Huawei, a Chinese network solution provider, have demonstrated remote control of a commercial car using a 5G network setup. In the demo a SAIC brand iGS car was controlled by a driver 30km from it. The driver used feeds from high-definition cameras mounted to the vehicle streamed over a 5G network providing him with a 360-degree view of the surroundings. Control signals for the brakes, steering wheel and accelerator were also transmitted over the 5G network with a latency of 10ms. Huawei claims the driver was able to fully control the vehicle at all times. Huawei sees 5G remote driving having applications for vehicle control in hazardous situations such as at mines, or to remotely control a fleet of vehicles.

LiDAR tricked

Researchers affiliated with the Korean Advanced Institute of Science and Technology says that they can confuse a Velodyne VLP-16 LiDAR system with two types of attack. The first attack the team conducted was a saturation attack that involved illuminating the LiDAR with strong light of the same wavelength as that the LiDAR uses, with a result similar to a human being blinded by strong sunlight. The second attack was a spoofing attack that involved creating an optical illusion for the LiDAR and making it appear that the illusion was closer to it than the object creating the illusion. The curved glass that is a feature of many LiDAR systems allowed the scientists to use refraction to change the point cloud that LiDAR uses to build a world picture. The team says that these attacks would be difficult to counter as it would involve multiple checking and crosschecking of all information from the LiDAR unit slowing the data processing and having a significant negative impact on the speed of autonomous cars.

Waymo says goodbye to Firefly

Waymo, Google’s self-driving car company, announced that its iconic Firefly self-driving car prototype is to be retired. Waymo plans to focus its efforts on integrating self-driving technology into Chrysler Pacifica minivans. The change is symptomatic of Waymo’s push to help create mass-produced self-driving cars. The fleet of 600 Pacifica minivans are equipped with the technology used on, and developed in, the Firefly including radar, LIDAR and vision systems, and a new AI compute platform.

Proposal for exemption for autonomous cars

Members of the Energy and Commerce Committee’s Digital Commerce and Consumer Protection Subcommittee in the USA have proposed legislation which would allow the testing of 100,000 autonomous cars on public roads throughout the country. The bill would exempt these cars from following the laws that state that autonomous cars must still have a steering wheel and other human controls.

PSA Groupe roadmap

French automotive company PSA Groupe – the maker of Peugeot, Vauxhall, and Citroen vehicles – has released its roadmap to autonomous cars. In 2018 the company plans to have automatic driving function under driver supervision – level 2 autonomy – in the DS 7 Crossback. The car will feature autonomous speed and trajectory control, self-parking, night vision for night driving, and driver attention monitoring. In 2020 PSA Groupe envisions a fully autonomous car without any driver supervision.

Add this: