Will Autonomous Vehicles Make Cyclists Safer?

As cyclists, my fellow two-wheeler's and I always wonder whether this ride will be the one where we get hit by a car or truck. It's inevitable they say. According to the U.S. Department of Transportation's Fatality Analysis Reporting System (FARS), a total of 854 cyclists were killed in crashes with motor vehicles in 2018. While deaths among cyclists younger than 20 have declined 89 percent since 1975, deaths among cyclists above 20 and older have tripled. Indeed, the popularity of cycling has added many more cyclists to our roads albeit at a cost of more accidents and lives lost as encounters between vehicles and cyclists become common in ever growing city populations. I have been cycling since my teens and consider myself an experienced cyclist, always being aware of my surroundings and monitoring the changing road situation for potentially dangerous. That being said, my fear of getting struck from behind or by a car pulling out in front of me is always my biggest concern. I can always spot a bad driver up ahead and adjust my path to avoid them or be prepared to act quickly just in case. I've bunny hopped many a curb onto sidewalks or grassy yards to avoid contact with vehicles in my 50+ yrs riding a bike. Also, a week doesn't go by when, even cycling in a dedicated bike lane, a passing car veers close to me as they pass noticing that they are texting or looking at their smartphone display. As an engineer with interests in the latest technologies, I've come to embrace the promises of increased safety that autonomous vehicles and their developers promote in their marketing materials and real life test reports. With technologies such as V2X (vehicle to everything communication), I strongly believe the deployment of this technology in autonomous vehicles will make our streets safer for all cyclists.

Vehicle to Everything Communication (V2X)

V2X is a vehicle communications system that passes information from the vehicle to any entity such as infrastructure (V2I), network (V2N), vehicle (V2V), pedestrian (V2P), device (V2D) and grid (V2G). There are two types of V2X communication technology being used, 1) WLAN-based and 2) Cellular based. While I won't get into the technical specifications here, WLAN-based V2X was developed first and supports direct communication between vehicles (V2V) and infrastructure (V2I) using Digital Short Range Communication (DSRC). Cellular based V2X (C-V2X) initially published specifications based on LTE but in its latest release, V2X functionality have been expanded to support 5G communications with the advantages being short latency and greater transmission speeds for real time response. Within vehicle to everything communications, the opportunity exists for vehicles to see bicycles and communicate awareness to other vehicles within the vicinity. Additionally, autonomous vehicles, upon detection of cyclists, will be able to react much faster than their human counterparts and insure the safety of both pedestrians and cyclists on the roads. Below I describe my worst case bicycle-to-vehicle (B2V) scenarios and how autonomous cars and V2X communications can help prevent accidents and make the roads safer for us all.

Cars Want To Get Where They Are Going

Even with dedicated bicycle lanes, cyclists are in danger of getting hit by negligent drivers. Figure 1 below shows two photos where a red truck is coming from the right, exiting a small shopping center-strip mall and about to enter the dedicated bike lane, and a small SUV coming in the opposite direction, turning into the shopping center entry.

Compounding the danger of this situation are pedestrians standing in the bike lane up ahead trying to cross to the other side. In my autonomous vehicle world, the SUV turning in front of me would have used their on-board sensors to perform image classification and identified me as a cyclist letting me pass safely instead of turning in front me. The use of LiDAR (Light Detection and Ranging) technology in autonomous vehicles can immediately tell the distance to an object and direction of that object. In today's real world, cars, trucks and SUVs always believe that they can squeeze in their turn before the cyclist which places the responsibility on the cyclist to always be analyzing the road in front of them. The red truck making a right turn onto the lane was preoccupied with the pedestrians in his path so I seriously doubt they were aware of a bicycle coming from their left down the dedicated bike path. Autonomous vehicles with their on-board cameras, LiDAR and AI-enabled computers will be able to make real-time calculations on the fly and react accordingly to whatever they are presented with. By knowing the distance and speed of the cyclists, autonomous vehicles would be able to calculate whether it was safe to turn in front of the cyclist. While most of today's automobile drivers believe that cyclists are slow moving objects, the introduction and popularity of e-bikes is going to shatter that myth quickly as e-bikes can easily go 28 mph or faster, depending on regulations. And as other fellow bicycle racers reading this know, we pedal pushers can get up to 30+ mph speeds on flat roads and have surprised many a car wanting to turn in front of us.

Figure 2 above shows a common dangerous situation for cyclists. How many times have we been cycling on the road, dedicated bike lane or not, and all of a sudden, you hear the car passing next to you, accelerate and turn in front of you. Not only does the driver of the vehicle underestimate how fast you are going but cutting in front of you could end up with the cyclist slamming into the back of the vehicle. While this scenario shows plenty of distance to slow and allow the vehicle to turn, many of us have been in situations where the driver of the vehicle didn't leave enough time for you to react and you ended up either slamming into the back of the car or bunny hopping the curb to avoid hitting them.

Autonomous vehicles passing you would have identified you, calculated your distance/speed to the intersection it wants to turn into and let you pass safely. A bicycle traveling at 23 mph would travel at 33.77 ft/s so that 100 foot distance to the intersection would take approximately 3 seconds. A passing car, traveling at 35 mph would travel at 51.33 ft/s so covering the 100 foot distance in about 2 secs leaving 1 whole second to slow, brake and turn safely in front of the cyclist which isn't adequate time at all. No human driver can do these calculations quickly except maybe Sheldon Cooper but he doesn't drive. An autonomous car can easily make these calculations in real time and decide to turn or wait for the cyclist to pass.

Good Old Country Roads

What cyclist, and even motorists, doesn't love a nice ride through an old country road with shaded trees, twisty turns and hilly ups and downs. Unfortunately for both cars and cyclists, these roads typically have zero shoulders or dedicated bike paths which are recipes for disaster. Sometimes fatally. We have quite a few of these roads in Austin, one of, if not the largest growing metropolitan in the country which means more vehicles and cyclists trying to share these roads on any beautiful weekend day. Figure 3 below shows another one of my nightmare scenarios to get hit by vehicles trying to pass me.

Typically, it's up to the passing vehicle's driver, in this case the blue SUV, to make a decision when to pass a cyclist riding on the right side, or left in you're in the UK or other Commonwealth nations. Unfortunately, this leaves it to the experience of the driver and frankly, most drivers aren't going to be able to adequately judge if they have enough time to move into the opposite lane, pass the cyclist and move back into their lane before the oncoming vehicle arrives. I couldn't tell you how many times I've seen vehicles almost clip each other head-on or worse, force me to bike off the side of the road onto the grass or dirt hoping my mountain bike skills will come in handy so I don't crash and injure myself. This scenario represents a excellent use case for autonomous vehicles to not only identify the cyclist, their traveling speed and distances between them but to also communicate with the approaching car using vehicle-to-vehicle (V2V) communications to receive their speed information, make real time calculations and safely pass or don't pass. While autonomous vehicles and V2X communications won't solve the issue of too many vehicles and bicyclists on the road, at least cyclists can rest assured they won't get run over or pushed off the road possibly injuring themselves if they fall off their bikes. Additionally, drivers of autonomous cars won't be grief stricken by having struck a cyclist, injuring them or worse, killing them.

My Autonomous Vehicle World

As I've mentioned before, I am a tech guy (BSEE from the University of Texas in Austin) having spent over 35+ years working in the semiconductor and software industries. As of late, I've been consulting in how Artificial Intelligence (Machine Learning-Deep Learning) and Computer Vision technologies apply to the transformation of city roads to Smart City ones. Smart Cities of the near feature will play a key role in addressing how autonomous vehicles interplay within the infrastructure all deploying amazing technologies such as V2X, AI image/object classification, LiDAR and 5G. Part of that infrastructure includes cyclists, pedestrians, scooters and other yet to be identified personal modes of transportation. Figure 4 below shows how autonomous cars will not only be communicating with each other via vehicle-to-vehicle (V2V) communications but also how their own onboard sensors and cameras will be scanning their environment to determine objects such as cyclists and pedestrians in their paths.

In our scary scenario from before, autonomous cars passing you from behind will be able to identify you and calculate to keep safe 3 feet passing distances required by law in some US States. V2X includes options for vehicle-to-bicycle communications assuming manufacturers of bicycles or bicycle equipment such as GPS-enable computers, smartphone apps and even helmets, integrate the required electronics and software to be compliant. Assuming our bicycle or other equipment does support V2X communications, it would be able to transmit (B2V) to the red vehicle trying to turn right that the cyclist is cruising down the dedicated bike path and to hold off and let it safely pass. Additionally, since the oncoming car trying to turn left into the shopping center may not be able to identify a cyclist on the opposite bike path as they are partially blocked by passing cars, vehicle-to-vehicle (V2V) communications would be able to notify the oncoming cars that it just passed a cyclist and to not turn immediately after the vehicles have passed. Autonomous cars would not turn in front of oncoming traffic but as it happens today, cyclists obscured behind passing cars may not be seen and the oncoming car turns left, right into the cyclist or cyclist's path.

While it may take 5, maybe 10 years for autonomous cars to be deployed everywhere in city grids, I'm hoping there is some effort on the part of bicycle manufacturers and equipment manufacturers to provide some early capability. Trek Bikes, one of the largest bicycle companies in the world has teamed with Ford and Tome Software to create a bicycle-to-vehicle communications system. You can read the Wired article here, https://www.theverge.com/2018/1/9/16870614/ford-trek-tome-bicycle-to-vehicle-communication-ces-2018. GPS-enabled bicycle computer companies such as Garmin and Wahoo, as well as any enterprising smartphone app developers out there, can get into this nascent market and develop next generation devices that communicate with the autonomous car revolution and Smart City transformations. So do I believe autonomous cars will make it safer for cyclists? The answer is yes! Yes they will.