Samsung Starts Development on Chips for Google’s Self-Driving Company

25-03-2021 | By Robin Mitchell

Recently, Samsung has begun development on producing core semiconductor products for Waymo. Who is Waymo, what is Samsung developing, and how will Google’s technology work?

Samsung Developing Semiconductors for Waymo

A recent report has stated that Samsung has begun development work on key semiconductors for Waymo. Self-driving vehicles will most likely be armed with an array of sensors including cameras, LiDAR, SONAR, and RADAR, and as such real-time processing is required. However, since a vehicle will need to interpret the road instead of simply detecting objects (i.e. unclear road markings, potential obstacles, and diversions), AI will be vital.

While the exact hardware being developed by Samsung is currently unknown, it will most likely be in the form of dedicated neural processors, GPUS, and state-of-the-art processors. Furthermore, the use of networking and cloud systems will also require high-speed, low-latency modems. 

A subsidiary of Samsung, Harman, is in the process of acquiring a startup company called Safari which specialises in self-driving cars. While this may suggest that Samsung is also aiming to produce self-driving vehicles, its Harman subsidiary specialises in automotive electronics, which will most likely be involved in developing semiconductors for Waymo.

Who is Waymo?

Waymo is a subsidiary of Google and is Google’s self-driving development company. The company was founded in 2009 as the Google Self-Driving Car Project but was later renamed Waymo in 2016. Waymo has raised $3 billion in total, which shows the potential value in self-driving systems. Waymo is currently the only company that has run a Robo-taxi service with no driver at the wheel (i.e. entirely self-driving).

Waymo has over 20 million miles in recorded self-driving tests and has been rigorously tested in many different locations. Furthermore, Waymo has a database on self-driving vehicles which is significantly larger than any of its competitors, putting Waymo in a leading position.

What technology will Waymo utilise?

According to Waymo, the self-driving cars utilise multiple navigation technologies to map the surrounding environment and predict what will happen. The first key piece of technology used by Waymo is LiDAR which allows the vehicle to create a 360 degree detailed map of the immediate area. The second technology used is RADAR which enables Waymo to determine ranges from objects and their speed.

High-resolution cameras are used to determine road signs, traffic lights and identify objects. Furthermore, the classification of objects (such as pedestrian and car) can help the system predict what those objects will do.

However, the report stating that Samsung is in the process of developing semiconductors, it also stated that Waymo systems will make use of Google Cloud Services. As such, we can make an educated guess on what network technology Waymo will utilise for connecting with Google Cloud Services by understanding the requirements of self-driving vehicles.

Waymo will most likely utilise internet services for training and updates instead of remote data processing. Since vehicles in motion require real-time data processing, decisions about the car, including direction and speed will be done using onboard systems. However, training the onboard AI will be done remotely at a data centre as data from other systems can be combined to provide a single training model with large amounts of data. 

Therefore, latency and connection speed is not critical. While technologies such as Starlink exist, it is more likely that such vehicles will utilise cellular networks. These are more established and simpler to implement (Starlink requires a phased-array antenna). Of all the cellular networks currently developed, 5G is the most likely candidate for Waymo vehicles, but not for its increased download speed and lower latency. 5G has the ability to support far more devices than 4G. If the majority of vehicles in the future become self-driving, there will be many thousands of simultaneously connected devices. As such, 5G would be a better candidate than 4G.


Read More

Profile.jpg

By Robin Mitchell

Robin Mitchell is an electronic engineer who has been involved in electronics since the age of 13. After completing a BEng at the University of Warwick, Robin moved into the field of online content creation, developing articles, news pieces, and projects aimed at professionals and makers alike. Currently, Robin runs a small electronics business, MitchElectronics, which produces educational kits and resources.