Courtesy of Robotic Research

Autonomous Operations

Autonomy is at the core of all good advanced robotic designs. What sets InstantEye apart from other systems is its mission-focused autonomy that enables extremely intuitive ease of use. Other systems may require the user to wait while the OS boots up, perform a pre-flight compass calibration, pair with Wi-Fi, or select communications channels. Many systems require the user to wait while the system acquires GPS or require the user to specify a GPS-enabled route ahead of time. We have worked hard to build autonomy into InstantEye that specifically doesn’t require any of these slow-downs and allows the user to launch within seconds from a stowed configuration.

In line with InstantEye Robotics’ design ethos, we are developing advanced autonomous algorithms for additional payloads and functionality that specifically target very low size, weight, power, and cost (SWaP-C) hardware. Ultimately, this allows the overall vehicle to be smaller, lighter and faster, which translates into better wind performance and longer flight times.

In particular, InstantEye Robotics is developing an autonomous operations payload with Simultaneous Localization and Mapping (SLAM) algorithms that will enable multiple InstantEye vehicles to enter, map, and return to the user, all without user intervention. This will permit autonomous mapping missions past line of site and radio link. The key behind this autonomy is a collection of custom-built algorithms specifically designed from the “hardware up” explicitly for low SWaP-C resources. The SLAM algorithm uses only the lightweight monocular camera and a small additional processor, relinquishing the requirement for heavy sensors like LIDAR, Xbox Kinect, and power-hungry processors. The inter-agent cooperative task assignment and multi-vehicle planning autonomy is designed using an auction algorithm that accommodates mapping and non-mapping tasks. Specifically, the loop-closing opportunities can be targeted and incorporated into the actual multi-vehicle planning, which empowers the group to truly be greater than the sum of its parts by planning towards targeted loop closing.  Algorithms were designed in simulation, validated on existing hardware, and are currently being implemented into the InstantEye framework.

Task assignment and cooperative mapping simulation (simulation midpoint)