Robots at CES 2018: Autonomous Vehicles

January 2018

Self-driving cars and autonomous vehicles was one of the most popular topics at CES this year, with a huge number of groups in attendance.

Aptiv / Delphi

They had 8 autonomous vehicles at CES and were giving ride-sharing demos in collaboration with Lyft.
Their vehicle is a BMW sedan, with 9 x LiDAR sensors and 10 x Radar sensors (Fig. 6).
Website: www.aptiv.com/ces-2018

Fig. 1:
Fig. 2:
Fig. 3:
Fig. 4:
Front sensors
Fig. 5:
Rear sensors
Fig. 6:
Top view

Navya Technologies SAS / Keolis Commuter Services

They were demonstrating their “Autonom Cab” which is a 6 seat autonomous taxi with no steering wheel, pedals or safety driver. They have also been conducting tests of their “Autonom Shuttle”, which is a 15 seat bus. The vehicles can be operated remotely by a web-based fleet management system.

The Autonom Cab has lots of sensors:
• 10 LiDARS (3 x 360° Velodyne VLP-16, 7 x 145° Valeo Scala)
• 6 Cameras
• 4 Radars

Website: www.navya.tech/en

Fig. 7:
Fig. 8:
Sensors on front of roof
Fig. 9:
Rear sensors

Roadstar.ai

This group is developing software for L4 autonomous driving. At their booth they had a Nissan SUV with a screen showing the output of their perception system, which was detecting and tracking humans (Fig. 15).

The vehicle has various sensors:
• 1 x Hesai 40-channel LiDAR (“PANDAR 40”)
• 2 x Robosense 16-channel LiDARS (“RS-LiDAR-16”)
• 2 x Velodyne 16-channel LiDARS (model unknown)
• 5 cameras (1 x forward-facing with some sort of wide-angle lens, 2 x forward-facing with lens caps attached, 2 x side-facing, 1 x rear-facing)

Website: www.roadstar.ai

Fig. 10:

Fig. 11:

Fig. 12:
Front view of roof-mounted sensors.

Fig. 13:
Side view of roof-mounted sensors.

Fig. 14:
Rear view of roof-mounted sensors.

Fig. 15:
Person-tracking demo.
You can see it misses two people,
but it does detect my waving hand.
Video 1:
Sensor fusion demo

Roadstar.ai
Video 2:
Panoramic sensor fusion

Roadstar.ai

Torc Robotics

They were demonstrating their autonomous Lexus SUV which is called “Asimov”.
It has one Velodyne HDL-64 LiDAR mounted high above the roof, so there is minimal blind spots close to the vehicle.
It has at least 6 cameras arranged to provide 360° vision and it looks like they’re using deep learning in their perception software.
Website: www.torc.ai

Fig. 16:

Torc
Fig. 17:

Torc

TRI (Toyota Research Institute)

They were showing their v3 vehicle for automated driving research.
It has a lot of sensors:
• 4 x Luminar LiDAR on the roof
• 4 x Velodyne VLP-16 around the lower body
• 10 x Radar, including custom sensors in the front fender (quarter panel), see Fig. 23.
• 9+ cameras, large-body Prosilica GT (4 x forward-facing, 2 x forward at 45°, 2 x side-facing, 1 x rear-facing)

Website: pressroom.toyota.com/CES

Fig. 18:

Fig. 19:

Fig. 20:
Side view of roof sensors.

Fig. 21:
Radar and LiDAR sensors in the
passenger-side fender panel.

Fig. 22:
Velodyne LiDAR in the front
bumper.
Fig. 23:
Location of Radar and
LiDAR sensors.

Toyota

Fig. 24:
Live output of fusion
of LiDAR sensors.
Fig. 25:
Close-up of custom triple-head
Radar sensor in the front fender.

Toyota
Fig. 25:
Installing the camera pod
on the roof.

Toyota

Elektrobit

This company is developing software for automated vehicles.
They had a fun demo at their booth, with some mini autonomous cars driving around a model of a city.
Each car had ultrasonic sensors around the bumper, a webcam and a depth sensor.
Website: www.elektrobit.com

Fig. 26:

Fig. 27:

Video 3:
Mini autonomous cars