When working on complex projects like autonomous tool detection and transportation in manufacturing environments, every algorithm, sensor, and piece of hardware counts. However, testing these elements directly on a large and sophisticated drone like Titan poses both a financial and operational risk. Enter Echo, our cost-effective quadcopter, serving as a crucial stepping stone in our algorithm testing pipeline.
Echo is a smaller, less expensive quadcopter but still packs enough punch to give us valuable insights. The lesser investment in Echo both in terms of size and cost means that we can afford to be more daring in our initial tests. Echo's primary role is to serve as the first real-world testing ground for all our algorithms before they are integrated into Titan. By scaling our algorithms to suit Echo's smaller frame and simpler hardware, we can conduct rigorous tests on control mechanisms, path planning, and collision detection.
So, while Titan might be the star of the show, Echo is the diligent understudy, waiting in the wings, ready to step in and help us perfect the performance.
Current Milestones
-
Completion of Echo's Construction: The construction phase of Echo has been successfully completed, marking a significant first step in our journey.
- Successful Manual and Pre-Planned Outdoor Missions: We have conducted manual and pre-planned outdoor missions, validating Echo's functional capabilities. The outcomes of these missions have been highly promising, laying a robust foundation for further advancements.
-
New Achievement - Indoor Preplanned Mission with Marvelmind Positioning System: Building on our previous successes, we recently achieved a major milestone by successfully executing an indoor preplanned mission using the Marvelmind indoor positioning system. This was a complex challenge, requiring specific adjustments to the flight control parameters to accurately interpret GPS data. One significant challenge we encountered was managing the update frequency of the GPS, which was crucial for having GPS-3D Fix on the Cube Flight Controller.
In the following video, we showcase this indoor preplanned mission. The mission objective for the drone was planned as taking off to a height of 1 meter, hovering for approximately 5 seconds, then moving horizontally to position itself above a designated right white paper. Subsequently, the drone maneuvered to the left, aligning itself above another white paper, before returning to its home position and executing a controlled landing. This demonstration not only highlights the drone's operational capabilities but also underscores our progress in achieving reliable indoor navigation.
The Road Ahead: Elevating Drone Capabilities with AI-Powered Object Detection
In our upcoming phase, we're set to enhance our drone's functionality by integrating a downward-facing camera linked to an NVIDIA Jetson Nano. This upgrade will harness the power of YOLO algorithm for real-time, precise object detection. Our focus will be on identifying various tools, a critical step towards autonomous tool delivery in industrial settings.
Once the system identifies the desired tool, the drone will autonomously navigate above it, descend, grasp it, and then transport it to a human collaborator. This advancement is not just about detection; it's about creating a seamless, efficient interaction between the drone and its environment, revolutionizing tasks in manufacturing and warehousing.