Nidhi Homey Parayil

Plants aren’t rigid or predictable. They bend, break, and hide what matters — and they don’t tell you when you’ve hurt them. My research explores how to model these properties so that robots can learn to interact safely and effectively. That includes:

• Physical interaction under occlusion
• Tactile sensing for leaf/branch contact
• Simulating deformable plants for robot training
• Vision + force integration for perception in the wild

I'm working to build plant-aware robotics that don’t just avoid nature, but learn to engage with it — across large-scale farms and small home gardens alike.

Worked on robotic pruning with Dr. Davidson as part of the AgAid Labour in collaboration with WSU. I worked on automating pruning tasks in sweet cherry orchards a labour-intensive seasonal activity currently facing severe workforce shortages. My contribution focused on developing a vision-based algorithm for robotic arms to follow "leader branches" in espalier-trained cherry trees.

We used an eye-in-hand camera configuration, where the camera was mounted on the robotic arm, enabling dynamic visual tracking along the branches. This continuous tracking was critical for locating valid pruning points and ensuring safe interaction without damaging surrounding plant structures.

The project culminated in a 5-day field trial on an operational farm, an eye-opening experience that highlighted the real-world gap between lab robotics and field performance. Exposure to unpredictable lighting and irregular plant geometry gave me an understanding of the need for robust perception and control in agricultural settings. Read Paper

even without leaves, navigating branch structure proved challenging. That experience strongly influenced my PhD direction: I realized how significantly occlusion from foliage complicates physical interaction, perception, and planning in agricultural environments. Understanding and solving that complexity became the core motivation for my current research.

At IISc Bangalore, I worked with Dr. Dinesh's lab to bring a custom-built six-axis robot to life for prenatal scanning. My role focused on integrating ROS with the robot’s CAN bus to enable motion control.

As someone from a production engineering background with limited coding experience, it was a steep but transformative learning curve. I also got my first exposure to modeling deformable objects, understanding how the robot might safely interact with the human abdomen. Though the project was exploratory, it introduced me to the complexity of soft-body interaction and sparked my current research focus.