Case Studies: System Development
CR’s robotic capabilities are broad and deep: perception, positioning, motion and path planning, power, computing, manipulation and mobility/platform design. Customers partner with us to apply one or more of these capabilities to solve a particular problem; to convert an existing platform into an autonomous system; or to create a custom-designed, mobile robot.
Standoff Robotic Explosive Hazard Detection (SREHD) - Neutralization
In September, 2014, the US Army awarded CR a $22M, multi-year Engineering, Manufacturing and Development (EMD) contract to develop the Autonomous Mine Detection System (AMDS). Now referred to as Standoff Robotic Explosive Hazard Detection (SREHD) - Neutralization, SREHD is a semi-autonomous mine and improvised explosive device (IED) detection system that provides the ability to remotely detect, mark, and neutralize buried, metallic and low metallic mines, bulk explosives, and various IED components and thus removes the soldier from the dangerous mission of searching for threats by hand. After detecting a device, the soldier directs the system to mark and neutralize the threat. The system is currently undergoing government testing.
SREHD includes a custom-designed, five degree-of-freedom manipulator arm that swings the AN/PSS 14 detection sensor across the width of the lane to be cleared. A CR-supplied MultiSense S21 stereo camera creates a 3D terrain profile to safeguard the sensor and guide movement of the arm. An intelligent payload package installed on top of the robot prime mover provides power, positioning and computing. The team designed the system to meet demanding US Army environmental, SWAP and reliability goals. CR will deliver a full technical data package including operator and maintenance manuals.
Nilfisk Autonomy Kit
CR is partnered with Nilfisk A/S to build advanced and intuitive autonomous floor cleaning robots. The Advance Liberty A50 is the first product that will result from this collaboration. The A50 is a stand-on scrubber/dryer that can be operated autonomously or manually and features two operating modes. In Copy Cat mode, the operator trains the autonomy system by cleaning a space manually. Later, the A50 can “replay” those paths autonomously, allowing the operator to perform other cleaning activities nearby. With Fill-in mode, the operator drives the room perimeter and the A50 automatically plans a path to clean the interior of the space. With its proprietary sensor suite and software, the scrubber/dryer recognizes unknown obstacles as small as a tennis ball and then automatically maneuvers around them, making the Advance Liberty A50 safe to use in open or congested spaces. The A50 will officially launch in the spring of 2017. The Nilfisk-CR strategic partnership will produce additional autonomous floor cleaning products in 2018 and beyond.
The A50’s autonomy is powered by CR-designed hardware and software optimized to operate Nilfisk scrubbers. Its multiple sensing modalities support different perception functions (positioning, obstacle detection and avoidance, drop-off and people detection, etc.) and autonomous operation in different interior spaces with varying size, floor, wall types and lighting conditions. CR applied a proven simultaneous, localization and mapping (SLAM) algorithm that underpins its industry-leading GPS-denied position estimation technology. CR will supply and support the sensors and embedded computers that form the autonomy kit to Nilfisk.
CR serves as the early phase technology arm for an agricultural startup, Rowbot Systems LLC. Rowbot is commercializing a concept to fertilize row crops with a fleet of small robots; the devices operating between crop rows to apply nitrogen fertilizer when the crop can most benefit. A single operator oversees the fleet and performs maintenance and routine support functions. Rowbot partnered with CR for access to a strong technical team that was able to build a fully operational concept demonstration system in under 8 months.
A 30 inch standard corn row spacing constrained the platform envelop and required CR to design a 22 inch wide, mid-body steered, custom vehicle that stores the liquid nitrogen in low tanks that double as the chassis structure. Powered by a 13 hp diesel engine, the system moves at 4 mph through the corn rows applying nitrogen at the plant base. A front-mounted MultiSense SL camera provides the 3D information to guide the system autonomously through the corn rows.
CR provides a high-speed (great than 36,000 plants per hour) computer-vision-based sorting product for strawberry seedlings to a California-based customer. This system must work in real-world farming conditions, with low-skill operators, on a very wide range of plants. CR transitioned advanced machine-learning algorithms, licensed from Carnegie Mellon's National Robotics Engineering Center, to produce a smart learning system which can be trained on-the-fly to adapt to changes in field conditions, plant varieties, and yields. CR also developed the ruggedized, high-performance cameras, lighting, and processing hardware for commercial deployment of multiple systems, and provides support and upgrades to these systems in the field.