According to the U.S. National Highway Traffic Safety Administration, human error accounts for 94 percent of car crashes. This triggered an opportunity for autonomous vehicles to possibly make driving safer and easier. Besides the giant full-stack manufacturers like Tesla, Waymo, and GM, autonomous-vehicle technology has also given birth to a whole new startup ecosystem — no-stack self-driving startups. These startups make hyper-detailed maps, super sensitive LiDARs, driver-assist software systems, behavior-analysis tools, cross-communication technologies, and so many other effective solutions for all the nitty-gritty of self-driving cars. Let’s take a look at seven such interesting startups.
1. Nauto: Making software to sound an alarm
Nauto makes software for autonomous cars. Starting an autonomous-vehicle tech company means, accepting that you won’t be generating revenue for a while. But Nauto is different as it generates income from its connected cameras (Dash Cam) and AI for non-autonomous cars.
These cameras can record footage inside and around the vehicle. This can be of a great help to commercial fleets and insurance companies.
Detecting human driver behavior will also help its software for autonomous vehicles, as those vehicles are going to share the roads with human drivers. With this in mind, Nauto launched its new product called Nauto Prevent. Nauto Prevent detects when drivers have their eyes off the road for more than five seconds and driving at 60 mph. Once detected, it alerts drivers with voice commands. If the driver ignores the voice commands, it rings an alarm. This way it prevents collisions. Hence the name, Nauto Prevent.
Partners and investors
In its last round of funding alone, Nauto raised $150 million from investors like Greylock and Softbank. The company also partnered with key players like BMW and GM. Also, Nauto is persistently strengthening its team by acquiring talent from the most successful companies. Recently, Nauto hired Jennifer Haroon (the former head of business of Waymo) and Sanket Akerkar (the vice president of global enterprise sales of Microsoft).
2. Comma.ai: Making software to mimic Tesla
George Hotz, known for unlocking iPhone and reverse engineering Sony PlayStation 3, founded Comma.ai. After working with tech giants like Google and Facebook, Hotz decided to develop self-driving software attempting to take on Tesla.
They rolled out an open-source driver assist system called OpenPilot software. Some features of OpenPilot like the automatic steering on highways seem to mimic Tesla’s autopilot. This can be installed in (supported) cars using Comma’s EON Dashcam DevKit. The system provides real-time road monitoring, and the panda on-board diagnostics (OBD) II dongle allows you to interface with vehicle diagnostics. Interestingly, the “classic” Tesla Model S without autopilot was recently tested with Comma.ai’s OpenPilot software. As a result, the non-Autopilot Tesla Model S was converted into a vehicle with Autopilot-like features.
OpenPilot: The Android for cars
Years ago, when all the phone manufacturers were trying to make their own OS, Apple rolled out iOS. iOS redefined what a phone can do. But iOS was only limited to Apple phones. Then came the open source Android enabling all phone manufacturers to use Android for free. The same thing is happening in autonomous vehicle industry now. Tesla rolled out Autopilot for its cars and Comma.ai is building OpenPilot for all cars. The company’s website calls OpenPilot the Android for cars.
Panda dongle: The Fitbit for cars
In July 2017, Comma.ai launched an $88 universal car interface called Panda. Panda is like Fitbit for a car. Panda plugs into a vehicle’s OBD port to collect and record driving data. Panda comes equipped with USB and WiFi capabilities. Also, Panda can charge your phone on the go. Panda was demonstrated with two other software: Chffr and Cabana. Chffr (pronounced "Shiffer") is an app that can record driving data. Cabana is a software used to interpret that data. Today, Comma.ai boasts more than four million miles driven without driver input.
3. Ouster: The LiDAR maker
Ouster makes LiDAR and perception technology for self-driving cars. LiDAR (light imaging, detection, and ranging) is a system used to calculate the distance of a car from a nearby object. LiDAR sensors work like RADAR, but instead of radio waves, they use laser light. The laser light used by LiDAR is invisible and these sensors have become an inevitable component of autonomous cars.
Ouster initially launched an OS-1 LiDAR sensor and then rolled out a slightly different budget-friendly OS-1-16. These sensors cost only $3,500 and were ideal for drone navigation and surveying applications. Recognizing the demand in the automotive industry, the company rolled out OS-2-64 exclusively for autonomous vehicles.
OS-2-64 — The ultimate LiDAR for self-driving cars
OS-2-64 delivers visibility over a 200-meter range and has 64 Laser beams in a tight spacing. This sensor calculates the car’s distance from nearby objects and helps it take immediate action. Collision detection is done several times a second. The sensors not only measure the distance from objects but also help the car’s computer recognize what type of object it is. For example, if the object is a human being or another vehicle, the car will honk before it applies the brake. If the object is static (say, a sign) the computer will steer the car away. OS-2 is slightly larger than OS-1.
Ouster vs. Velodyne
Velodyne, a company that’s quadrupling LiDAR production, is the chief competitor of Ouster. However, Velodyne’s prices remain out-of-reach for most consumer applications. Ouster is trying to outdo its competitors by making affordable sensors. The company’s mission is to push LiDAR sensors to every consumer automobile. Ouster seems to have set a very aggressive timeline to scale out its LiDAR manufacturing, but the initial results are impressive.
4. Lvl5: Hyper-detailed map maker
Experts believe that LiDAR is just not enough to ensure safety. Hyper-detailed maps are necessary for self-driving cars in addition to sensors like LiDAR. Sensors can tell the car there is an obstacle, but it is important to have some kind of backup when the sensors fail. For example, in 2016, a car owner died when riding in a Tesla in Autopilot mode. The car mistook a tractor for a bridge and concluded that the car can pass under it. Considering a scenario like this, if the car had an updated HD map, it would have known that there was no bridge there. This is why Lvl5 sees the need to create HD maps.
How are HD maps different from conventional digital maps?
HD maps are way more precise than navigation maps having a level of accuracy down to 10 cm. This kind of accuracy can help vehicles make precise turns. Lvl5 collects mapping data from drivers of ride-hailing companies like Uber and Lyft. Lvl5 pays drivers through an app called Payver. Payver collects video footage of roads and other data from GPS, gyroscopes, and accelerometers from drivers’ phones and uploads them for Lvl5 to process. Lvl5 uses this data to create high-definition 3D maps.
In just three months, Lvl5 has recruited 2,500 drivers and mapped 500,000 miles. The company says that it would need 50,000 Payver users or just a partnership with one carmaker who would install their software in all their cars to collect all the data it needs to make its HD maps. The process does not stop with creating maps but Lvl5 will also need to update them consistently as self-driving vehicles need real-time information about things like road construction, accidents, natural calamities, and busted traffic lights.
5. Autotalks: V2X chipset manufacturer
While other companies are trying to make cars understand traffic, Autotalks is making vehicles talk to each other. Autotalks manufactures V2X (Vehicles to others) chipsets. V2X is a cross communication technology that allows smart vehicles to communicate not just with another vehicle but with the entire local infrastructure. The chipsets can enable cars to even recognize potholes and inform the municipality. This way the V2X technology ensures safety not just in terms of crashes but road maintenance as well.
The V2X bandwagon
Hyundai recently partnered with Autotalks to help the startup bring this technology to the market. Volkswagen plans to jump on the V2X bandwagon by 2019 and Jaguar Land Rover seems to have an eye on the technology as well. Toyota announced that it will be adopting cross communication technology by mid-2020s.
6. Zendrive: The driving analytics app maker
No matter how well autonomous vehicles are equipped with software, maps, and sensors, they will need to know about human behavior to avoid accidents. Especially when it comes to level 4 (cars that require human assistance at critical conditions), driving behavior analysis is required for setting safety standards and vehicle insurance. This is why Zendrive rolled out its driving-analytics app.
The Zendrive data-analytics app gathers data about the driver’s behavior using the accelerometer and gyroscope in the driver’s phone. The company has good chances of generating revenue by selling its analytics data to various companies that make machine learning software for autonomous cars.
Driver protect with Life 360
Zendrive has also put the data to use by partnering with Life 360 and launched an app called Driver protect. The app detects when a Life 360 user is involved in an accident and immediately informs the driver’s family, a spontaneous response team, and the local authorities. The beta program made this feature available to 50,000 users. Zendrive also enables parents to get driving reports of their teen drivers for which they will have to pay a subscription fee of $7.99 per month.
The company conducted a study in 2017 that revealed 88 percent of drivers were on their smartphones for an average of 3.5 minutes every hour. According to a study by Oregon state university, taking your eyes off the road for two seconds increases the chances of a crash by 20 times. Knowing that they are being watched will make drivers be careful with their driving behavior.
Self-driving cars in the fast lane
The world is inching toward autonomous technology day by day. The tech will, for sure, make driving safer, smarter, efficient, and regulatory compliant. But what will be the other consequences? Will drivers become obsolete? Will electric vehicles push fuel prices down? Will steering wheel manufacturers go bankrupt? Will kids have their own cars? It’s an uncertain and yet exciting road ahead for self-driving cars and the technology that will make them go.
Featured image: Shutterstock