Mike Potts (left) and Mark Preston, founders of StreetDrone: “We are helping to accelerate development of autonomous vehicles.” (StreetDrone)
StreetDrone: rolling R&D vehicle for small-scale AV developers
View gallery »
Even for well-funded automakers, suppliers, start-up businesses and academic institutions, the cost of developing specific technologies needed to support autonomous-vehicle (AV) R&D can be daunting. United Kingdom start-up StreetDrone, however, has devised a potential cost shortcut to the self-driving development cycle.
The company has developed what it regards as a competitively-priced “open-platform” version of the Renault Twizy electric vehicle (EV), fitted with “generic” autonomous-driving hardware, onto which can be grafted almost any specific CAV (connected autonomous vehicle) operating system and other electronic- or software-development technology, obviating the need for “reverse engineering, hacking or voltage interceptors” of a conventional production vehicle.
StreetDrone also is promoting the use of open-source software to help speed advanced vehicle development and share cost.
Merging racing with autonomy
Early in 2017, Mark Preston, who previously worked with McLaren, Arrows and Super Aguri and is now Team Principal of Techeetah Formula E, joined Mike Potts, an entrepreneur specializing in smart businesses, to form StreetDrone (its first vehicle is simply named StreetDrone ONE) to create what is essentially a mobile, targeted development-and-test platform. He told Automotive Engineering: “We are helping to accelerate development of autonomous vehicles and to make that facility available to any business or organization—not just large companies—and speed up autonomous vehicle development in general.”
The StreetDrone ONE’s specification includes two ECUs forming a 2-channel safety system; an arm-and-engage mode-selection system; a passive data-acquisition mode; drive-by-wire e-stop with optional passenger extension; a full-time safety-driver override system and two accessible CAN bus interfaces.
Simple and flexible platform
Preston and Potts chose the little, tandem 2-seat Twizy partly because of its modular construction—which includes clip-on plastic body panels on a spaceframe chassis—but also because it provides a relatively inexpensive, off-the-shelf solution: adding development and testing equipment is easy, stated Preston. Moreover, the Twizy’s ECU is not as complex as those of most “regular” passenger vehicles, so the CAN bus is not receiving messages from minor systems such as climate control that may not be germane to an autonomous-related system under development.
“A vision-systems or Lidar developer doesn’t need to be expert regarding the whole vehicle,” said Preston.
Potts added that many companies now involved in self-driving development “couldn’t do the car bit” on which to develop them: “If you are a software engineer you don’t have to know how to build a complete autonomous car—we can provide it. We are facilitating testing of very distinct products: computers, sensors, cameras (StreetDrone ONE integrates eight cameras as standard, all linked to an Nvidia Drive PX2 graphics processor), to test them in the real world. The car itself is just a small part of the product set, but is essential for real-world development of those specific self-driving systems fitted to it.”
The first StreetDrone ONE was delivered in March 2018. Five have now been completed, fitted with a pack of ECUs that the company designates XCU, to provide the link between the vehicle and self-driving hardware such as the PX2.
StreetDrone uses a PX2 to host its own open-source software, Open SD—a self-driving “stack” based on Autoware software, but any type of software (proprietary or open-source) can be installed, the developers said.
StreetDrone ONE also could be used in its own right as a “last-mile” delivery vehicle or the driver’s seat could face backward to create a taxi.
To create the StreetDrone ONE, the Twizy’s standard body panels are removed and replaced with custom bodywork to support sensors, hardware and lidar sensors. StreetDrone works closely with Renault as part of the POM (Platform Open Mind) program.
If a code being tested for any specific system does not operate as projected, the driver can assume control of the test vehicle.
Potts said, “Our objective is to help developers and others to become involved with the autonomous world. Getting to autonomy faster will save lives and will be fantastic for the environment. Economically, it is also a sound thing to do.”
He added that massive auto and software companies shouldn’t have monopolies in developing AVs: “There are a lot of people out there who have fantastic ideas and we think that involving them in the development of autonomous software and hardware will get the world to autonomy faster.”
The starting price (in British sterling) of the StreetDrone ONE is £59,500 and rises to £69,500 if fully loaded with the Nvidia Drive PX2 self-driving hardware.
Continue reading »