The client sought out DMC to develop the object detection algorithm, which needed to work in a variety of different environments such as varying lighting conditions and backgrounds. Because of our expertise with vision systems and our ability to package this solution at a low cost and production-ready design, DMC was a great fit for this project.
Our team reviewed example footage that was given to us directly from the client. We prototyped a detection algorithm as a proof of concept using NI vision processing libraries.
DMC’s embedded engineers ported that algorithm to the open-source OpenCV framework and chose a camera sensor and lens optimized for the environment of the robot. We optimized the algorithm to run a low cost, less capable embedded processor, a Raspberry Pi compute. We implemented a very specific memory and processing operation to have the algorithm run efficiently on the low-cost processor.
We then used the data from that detection algorithm to create navigation commands for the autonomous robot and tested the fully integrated system in the field, under real-world conditions. DMC collaborated with the client’s team to develop an exhaustive environmental test and perform field testing to gather more data.
Further, DMC went above and beyond the customers’ requirements by developing a diagnostic .NET desktop application to allow for visibility into the robot’s performance and further optimization from the client.
Learn more about DMC’s Embedded Vision Solutions. Contact us today to get started on your project.