����ֱ��

With rapid advancements in robotics and AI, the line between science fiction and reality continues to blur. At the heart of this innovation lies a breakthrough: drones designed to solve pressing global challenges, from pollinating crops to navigating wildfire zones.

This vision drives Assistant Professor Chahat Singh, leader of the�� (Perception, Robotics, AI and Sensing) Lab in the Paul M. Rady Department of Mechanical Engineering. With an academic background spanning electronics, robotics, and computer science, Singh is dedicated to exploring the frontiers of bio-inspired robotics and AI in resource-constrained systems.

��

Assistant Professor Chahat Singh next to one of his compact and autonomous robotic designs.

Singh’s overarching research question is deceptively simple: What is the minimum amount of computational power, sensor capability, and resources required for small robots to achieve autonomy? This challenge is compounded by the scale of the robots he designs, which are constrained by limited computational capacity and lightweight requirements. They are two to three inches in length and orders of magnitude smaller in terms of physical size and computational power than traditional robots. “We’re working with systems that have 100 times less computing power than a Boston Dynamics’ Spot robot,” Singh explained. “The goal is to achieve autonomy with the bare minimum.”

One of Singh’s most notable projects focuses on autonomous drones for pollination, inspired by the overwhelming loss of honeybee colonies. “The question was whether today’s robotics and AI could fill this gap until we have a more sustainable biological solution,” Singh said. The answer lies in his innovative, lightweight drones that can navigate autonomously through forests and fields without relying on external communication or GPS, making them secure and efficient.

Singh’s current drone model incorporates multiple onboard cameras, which enables it to identify and align with flowers for pollination. The cameras use advanced neural depth-perception algorithms powered by AI-accelerated computers. Many creatures have developed different pupil shapes based on their habitats which allow variations in incoming light and amount of blur to help them determine the depth of objects. “The cameras are inspired by biological systems,” he explained.����

��

Singh showcasing the small scale of materials in his robot's design. His goal is to develop autonomous drones with less resources and power than traditional robots.

Singh’s drones are not just technologically advanced—they’re engineering marvels. Built from carbon fiber frames, these drones are lightweight yet robust, weighing around 250 grams. They use lithium ion batteries which are heavy and tend to die quickly, so he has started to look at ways to charge the batteries while the robots are outside.��

To overcome these limitations, Singh has developed a “mother drone” system. The larger drone carries smaller drones to the target area and acts as a mobile charging station. Once deployed, the smaller drones autonomously search for flowers and begin pollination. This approach not only extends operational time but also reduces the energy expenditure of individual drones. “It’s a highly efficient system that mirrors natural ecosystems,” Singh said.

While the pollination drones have gathered attention, Singh’s research has broader implications. His team is working on compressing advanced AI models, such as language and vision models, to operate on resource-constrained systems. “Imagine a robot navigating a forest during a wildfire,” Singh said. “It needs to make decisions on the spot, without internet access or pre-programmed instructions. That’s the next frontier—embedding foundational AI models into small, autonomous robots.”

Singh’s vision extends to deploying fleets of robots for tasks like firefighting, disaster response, and ecological monitoring. By creating swarms of cost-effective, autonomous robots, he aims to revolutionize industries that rely on expensive, large-scale systems. “Smaller robots are not just cool—they’re necessary,” he emphasized. “They offer safety, robustness, and cost-effectiveness.”

Despite the groundbreaking nature of his work, he is committed to open-source principles. “I believe in openness because this research is for the greater good,” he said. Singh has already shared software for drone operation and plans to release additional resources to empower other researchers and innovators.

When asked about his favorite part of the research, Singh highlighted the hope it brings for the future. “Whether it’s addressing ecological crises or enhancing technology, I want to create robot systems that are safe, innovative and sustainable,” he said. “This is about pushing the boundaries of what’s possible while respecting the natural world.”