Civil Maps, a developer of cognition systems for autonomous vehicles, announced the release of the Atlas DevKit platform, a hardware and software offering that enables the low-cost, real-time creation and conversion of sensor data into 3D semantic map data.
The reference kit, a car-mounted hardware unit and its companion software suite, is designed for self-driving car developers and others seeking to affordably build as well as utilize high definition, machine readable maps that can be crowdsourced. The hardware device is comprised of third-party sensors such as LiDAR, cameras, measurement systems, and communication devices in a self-contained, plug-and-play package. An alternative version of the platform, the Atlas Lite DevKit, integrates with a vehicle’s existing sensors.
“Advanced localization, map creation, and crowdsourcing of maps are key challenges facing those hoping to test and deploy autonomous vehicle technology,” said Sravan Puttagunta, Co-founder and CEO of Civil Maps. “The Atlas DevKit platform accelerates the pace of innovation by enabling developers to quickly and economically localize vehicles, build dynamic maps, and crowdsource that critical information with other cars in real-time.”
Affordable Data Collection and Localization at Scale
Traditionally, low-level localization and data collection can incur hardware costs alone in the $100,000-$500,000 range, while requiring significant computing power. Civil Maps is able to offer the full Atlas DevKit platform to qualified customers in packages that start at $20,000 as part of an R&D contract with the company. Highly efficient, the reference platform runs via a single board ARM processor, a remarkable advantage over other data collection and mapping solutions.
Since Fall 2016, automotive R&D teams have used the Atlas DevKit experimentally across three continents. Installation and usage is easy; developers simply mount the kit to a car’s roof rack and use it with Civil Maps’ machine learning software. Rather than utilizing expensive, custom-built solutions that can take months to deploy, R&D teams are able to start mapping immediately.
“In developing the Atlas DevKit, our goal has always been to reduce the costs of quality data collection, which is the pathway for crowdsourcing map information,” said Richard Hurlock, Product Manager for Atlas DevKit. “Now and moving onward with future iterations, we are developing with those scalable goals in mind, while providing real-time feature extraction and building maps that significantly outperform competing solutions.”
A key breakthrough on its own, the Atlas DevKit used with Civil Maps software, enables a car to localize itself in six dimensions within 10 centimeter accuracy. When combined with Civil Maps HD 3D semantic maps, the car not only knows where it is, but it also gains the ability to anticipate and remember objects in the physical world.
Highlights of the Atlas DevKit for Developers
● Advanced Localization: In 6 dimensions (x, y, z, roll, pitch, yaw) within 10 cm accuracy
● Affordable: Enables R&D at scale
● Minimal Compute: Runs on an ARM processor with low power consumption
● Low Data Footprint: (+/- 120 kb per km)
● Multi-Sensor Compatible: Integrates with a variety of sensors, allowing for interchangeable configurations
● Connected: 4G allows for remote provisioning
● Flexible: Atlas DevKit, Atlas Lite DevKit (for integration with vehicle’s existing sensors, Atlas DevKit SDK (for integration in hardware ready vehicles)
The Atlas DevKit was recently recognized as an Innovation Honoree in the Vehicle Intelligence category of the 2017 CES® Innovation Awards.