Mentors: Dr Girish Chowdhary & Dr Julia Hockenmaier

1. Building a voice-controlled system to enable users to control a CNC based gardening robot called FarmBot remotely. We are using the Alexa API to make custom Python and Lua scripts for interfacing with the web based application of FarmBot in real-time.

2. Working on integrating a pneumatically controlled soft robotic arm with the FarmBot to enable it to clear obstacles and harvest fruits.

Mentor: Dr Pratap Tokekar

1. Processed point clouds of the pasture obtained from the gazebo simulation for selected days of a yearusing LiDAR mounted on the hector quadcopter controlled using an autonomous navigation script.

2. Automated the task of constructing gazebo worlds for grass pastures where each plant has a unique pose and is scaled to match real world height data of a location.

3. Our work for long-term Spatio-temporal prediction of pastures using deep learning was published in Agronomy. We used an alternative approach for forecasting long-term pasture terrains using computer vision techniques inspired by U-Net and Monte Carlo Dropout inference methods for uncertainty estimations on historical pasture data measured using LIDAR.

4. Another publication using a multi-robot deployment policy for pasture monitoring by using the predictions from spatiotemporal deep learning is under review in IEEE Transactions on Automation Science and Engineering.

[Report] | [Project Code]

Mentor: Dr Simarjeet Saini

1. Developed an orange sweetness detector which used scaled conjugate gradient backpropagation in MATLAB to non-destructively predict sweetness of oranges using Near-Infrared spectroscopic data with an accuracy of 70%. Discrete cosine transform was used to reduce dimensions of input matrix and prevent memory overflow.

2. Designed prototypes in Solidworks, which were 3D printed using the Ultimaker 3. Developed low-cost photonic devices like Urea in milk detector and the Fundus eye camera. The Fundus eye camera used Raspberry Pi3 to capture images and videos of the retina of a model eye to aid in diagnosis of diseases.

3. The Fundus eye camera was featured in an article in the Optics and Photonics News (OPN) while the orange sweetness detector was showcased in a conference presentation for sweetness prediction using Support Vector Machine as the discriminant model along with Deep Q learning to tune its parameters.

[Report]

Mentor: Dr Dharmendra Singh

1. Varied the thickness of radio wave absorbers such as nickel ferrite on a UAV model and its individual parts like cylindrical body, spherical nose and cuboidal wings and detected its effect on the Radar Cross Section in Ansys HFSS.

[Report]