Autonomous driving technology relies on cars being connected. Managing the connectivity is going to be increasingly important...
Autonomous vehicle technology is expected to greatly impact connected cars, as much more data will come from an AV, and much more data and content will transfer to each AV. The graphic below shows my perspectives of how AVs will impact connected car data flow. It is a big-picture view showing key connected car segments. The figure looks at the major AV use-cases (even if the deployment timing has large uncertainties). The timeline at the bottom of the figure I believe is optimistic.
The figure uses color-coding for each of the five AV use-cases. ADAS, including L2 and L3, is in black. Goods AVs are shown in light blue, with robotaxis in dark blue . Autonomous trucks are shown in green. Fixed route AVs are shown in dark red and personal AVs have a light red color. A few questions are listed in two blocks with yellow at the right side of the figure.
This is the third in a series of Egil’s Eye columns on connected cars. The first two are:
Connected Car Industry: Big Picture (Part 1)
Connected Cars: Where’s the Money? (Part 2)
Each category has two blocks with information — one for data to the car and one for data from the car. There is a lot of data in the blocks that will be explained below. As expected, the AV use-cases will have similar impact since they use comparable AV technologies.
The lower levels of AVs or ADAS functions are already starting to make some impact on connected car capacity. Cybersecurity is more important to cars with ADAS functionality at L2 and L3 AV capability with resulting increase in connectivity volume. OTA (over-the-air) software updates also become more important for ADAS vehicles as the maturity of ADAS functionality needs considerable improvements based on AAA testing. Functional software updates are emerging for ADAS vehicles and will see future growth. This indicates more frequent software updates for ADAS cars than other ECU software updates.
Data from ADAS vehicles is also expected to increase. Road and map information updates are needed for L2 and L3 vehicles. Mobileye’s Road Experience Management (REM) and Roadbook are growing in importance.
With lots of uncertainty on how drivers use ADAS capabilities, it is important to gather and analyze ADAS feature usage patterns. This is another factor likely to increase the frequency of OTA software updates.
V2X (vehicle-to-everything) is likely to emerge for ADAS vehicles despite the battle between V2X-DSRC and C-V2X, the two competing methods for connecting vehicles. Timing and deployment speed is uncertain, but it V2X will add to connected car data usage.
The Covid pandemic impact is making goods AV the first segment for deployment among the AV use-cases. There are two types of goods AVs—sidewalk AVs and road goods AVs. With no people in goods AVs many of the traditional connected car functions are not needed. But the ECUs and specifically the ECUs controlling AV driving functions will need extensive connectivity.
This means OTA software updates and functional software updates will be very important. Goods AVs will primarily be used by fleet operators where the AVs return to a home base on a regular basis—usually multiple times per day. This allows for a reasonable secure update at the AV home base. Wireless connectivity will still be the dominant connectivity method, but wired connections will be a viable option for some functions.
Cybersecurity will be very important and will be a focus for all goods AV fleet operators.
AV operational data will be the most important data coming from goods AVs. AV operational data will be the key to improve the AV software and hardware through extensive analysis.
Goods AVs will rely on map and road data for navigation and will need up-to-date data. Hence changes in road data will be collected by goods AV for updates and distribution as needed to AV fleets.
The status of on-board goods and delivery timing will be crucial to the delivery customers.
Fixed Route AVs
Fixed and flexible route AVs have been impacted negatively by the Covid pandemic as shared rides are not desired. Fixed route AV potential include closed venue trips and a variety of bus routes. The fixed-route AV opportunities should be back to normal (or some kind of new normal) in a year or two.
The important connected car functionality for fixed/flexible route AVs are similar to goods AVs. The ECUs controlling AV driving functions will need strong connectivity as OTA software updates and functional software improvements will happen more often than currently.
Cybersecurity importance increases for all AVs with people onboard and will include cybersecurity software updates, extensive onboard software activity tracking to identify suspicious events.
AV operational data will be the most important data coming from route AVs and will be used to improve the safety and functionality of AV software and hardware.
The users of route AVs will require real-time data on availability and timing and this data will come from each route AV. Such data accuracy will determine the customer satisfaction and usage frequency.
Road data changes will be collected but is less important due to limited roads travelled.
The Covid pandemic created a large demand for trucking services that is helping to fund investments in autonomous trucking. Autonomous trucks have several sub-segments such as hub-to-hub trucking, closed venue trucks and fixed route trucking.
Connectivity capabilities are essential for autonomous trucks. The AV driving ECUs require wireless connectivity as OTA software updates and functional software improvements will happen frequently.
Cybersecurity at multiple levels is required for all AVs — especially for autonomous trucks considering the damage that could be done by successful black-hat hackers. Most autonomous trucks will be part of an AV fleet with a variety of cybersecurity defenses—from wide-ranging onboard software activity tracking to cloud-based cybersecurity operations center.
Again, AV operational data will be of great importance coming for autonomous trucks and will be a key to advance the safety and functionality of AV software and hardware.
Additional information and analysis from fleets of autonomous trucks will be needed to track AV safety improvement trends over many years. This will be required for all AV use-cases but are especially needed for AVs that replace human-driven vehicles. This data is necessary to establish and prove that AVs are safer than human-driven vehicles — how much, which driving environments and similar data.
Cargo data and status for trucks are already important but will become essential in the AV-based supply-chains of the future. More detailed data will be needed with higher frequency.
Most robotaxi testing and trials were halted in the beginning of the Covid pandemic with further impact and delays expected. Nearly all robotaxi trials use a safety driver. Waymo’s service in Phoenix use about 5% of their ride-hailing trips without safety driver. California regulation now allow AVs without safety driver, but tele-operation is then required. China is also starting to allow AVs without safety driver in some areas.
Robotaxis require larger wireless capabilities than the AV use-cases already discussed. Again, the ECUs managing AV driving require high-capacity wireless connections for frequent OTA software updates and functional software improvements.
Extensive cybersecurity capabilities are required for all robotaxis. Most robotaxis will be part of a ride-hailing fleet with onboard hardware and software cybersecurity defenses. Cloud-based cybersecurity operations center will track, analyze and compare connectivity activities for unusual patterns that requires investigations and actions when needed.
Robotaxis is expected to offer a large variety of content to its customers including video content that are not allowed for drivers. This is often called “passenger economy” to indicate that drivers become passengers and consume video-centric content. Any time video is used, the bandwidth requirements increase tremendously, and this will happen in robotaxis after volume deployment starts.
Robotaxis’ operational data will be especially important and can be leveraged from the variety of data that will be available from large fleets. Extensive analysis of this data will be a key to advance the safety and functionality of robotaxi software and hardware.
Additional information and analysis from fleets of robotaxis will be used to track how robotaxi safety will improvement over decades. Such data and analysis are needed to prove that robotaxis will be safer than current ride-hailing services with human drivers. It is important to establish which driving environments are best suited for robotaxis. What data to track and how it is used will require industry standards for fair comparisons and results that can advance long-term AV safety.
Robotaxi test vehicles are already collecting road and map information changes and this will continue as large deployment happens.
The passenger economy will increase the need for statistics on content consumption in robotaxis. A key question is how much privacy will be controlled by the user.
Personal L4 AVs are far away and have lots of uncertainties on future evolution and deployment timing. The best-case scenario is that personal AVs will leverage the robotaxi software driver platforms and could serve a similar area for personal use. This means that the connected car features needed for personal AVs will be similar to robotaxis as described above.
The passenger economy for personal AVs will be stronger than for robotaxis. This happens because personal AV passengers will spend more time in their vehicles, which should result in higher content consumption and more wireless bandwidth usage.
Other AV issues
There are a few questions on other connectivity issues that are listed in the two yellow blocks in the figure. This is focused on issues that impact wireless communication capabilities of AVs.
Tele-operation of AVs is a technology that will be used for some AV use-cases. This means that an AV will be controlled remotely by an operator that will see what AV sensors see. This will primarily be used for short time periods for specific events or as backup for to AV software issues. Teleoperation will require substantial wireless bandwidth to relay the sensor information to the teleoperator.
Wireless communication is crucial for all AVs, which creates a key question. Do some AVs require a redundant communication link to assure connectivity at all times? A similar question is whether the cybersecurity system needs its own wireless link to increase security. I believe the answer is yes to the second question. The extra wireless link can be used as the redundant link when needed.
Nearly all vehicles sold in the U.S. has a black box that records key ECU information when a crash occurs. However, this data is difficult and expensive to recover. I believe that AV crash data should be stored in a black box, but also transmitted immediately to a neutral party for storage and analysis. Such data can be transmitted over existing wireless connections.
V2X is coming to the auto industry even if we still have a technology battle between V2X-DSRC and C-V2X. AVs do not require V2X but the synergy and advantages of V2X are very promising. V2X will be another wireless link to non-AVs and AVs with volume deployment in 3-5 years.
Traffic flow data can be very useful to any vehicle and provide additional advantages to AVs that are even better with V2X included. Is it possible or likely that AVs will interact with future traffic management systems—at least on a local or regional level? If so, is this activity 10 or 15 years in the future?
Finally, a few more questions that will increase AV connectivity. First question is when will we have AV standards, how many and at what levels? (See: AV Safety Standards Set to Arrive in 2020)
Smart cities use data from current connected cars, but much more data from most AV use-cases will be very useful and will provide results that are beneficial to all vehicles. How much data is needed and when will the data sharing happen?
AVs creates tremendous amount of data—especially from the many lidar, radar and camera sensors. Most of the sensor data is short lived and will never leave the AV. The question is what portion of the sensor data will be transmitted from the AV to a cloud location? My view is that the portion is very small, but the total volume will still be big.
This creates a follow-up question. Data memory is often used as a trade-off to wireless transmission due to cost issues. The data memory is emptied later over a wired or local wireless connection. I think local data memory in the AV will be used for temporary storage for much of the sensor data.