VNIT students develop drone tech sans GPS

Nishita Gupta
Nishita Gupta July 31, 2023
Updated 2023/07/31 at 6:25 AM

Nagpur: Two students from the mechanical engineering department of VNIT developed an innovative model to guide drones in places where GPS signals are not available like inside buildings or complicated structures as part of their final-year project.

This model has real-life uses like in exploration of hard-to-reach places, helping with inspections, surveys, and even search and rescue missions. The simplicity and efficiency of this technology promise to benefit various industries, they said.

The BTech final-year project for the academic year 2022-23 was done by Araveti Aswartha Narayana Vinay and Utkarsh Vats under professor Shital S Chiddarwar.
The students selected a DJI Ryze Tello Minidrone — a mini drone — and used its onboard camera to map the route through visuals recorded by it. They called it “visual odometry,” a method to help the drone figure out its exact position and movement by carefully examining what it ‘sees’ in its surroundings, said Utkarsh.

“We employed advanced machine learning (ML) techniques. During the trial phase, we gathered data using special tools and the mini drone hardware. This data included important information about the drone’s location. They we trained different computer models like the ones used in recognizing images and patterns (CNN models), memory-based models (LSTM models), and models that could handle varying sequences of data,” said Vinay.
The innovation succeeded when they combined the strengths of these models. “We used the CNN models to understand what the drone’s camera saw and then passed that knowledge to the LSTM model. The LSTM model made predictions about where the drone was at different moments, without the need for complicated image processing,” said Utkarsh.

For more such content, keep reading @techinnews

Share this Article