Unique designed palm-size robot for swarm search application with long distance communication capability
Optimized morphology of the robot frame to minimize the size
Open-source sensor and software flexibility for research purpose
MATLAB Simulation of 5 swarm robots
Simulation of 10 Swarm robots in action
Virtual Environment to train and test swarm intelligent algorithm
Swarm Intelligence Algorithm for Search Application
A decentralized swarm intelligence algorithm for search application of source with spatially varying signal
This SI (Swarm Intelligence) Algorithm is derived from nature inspired Heuristic particle swarm optimization method
The Optimized SI Algorithm have high success rate of 78% with 5 robots and >98% with 10 robots to locate the source in unknown cluttered environment
Flying UAV Detection-Tracking System
DARPA funded project, objective is to detect and track a flying UAV using partially known physics based model along with data driven machine learning algorithm
For training ML model, a large data set of flying UAV in motion capture system collected with sensor fused data from multiple senor :LIDAR, Stereo-vision, Kinect, Thermal and RGB camera according to a well designed DOE
A trained hybrid model can predict to track UAV using partial physics model and various sensor inputs with low computational power
Comparison of Simulation and Testing (Fast Forward X16)
Behavior Trees for Obstacle Avoidance policies
Virtual Environment to train and test obstacle avoidance algorithm
Collision Avoidance Algorithm using Behaviour Trees
A behaviour trees based obstacle avoidance algorithm with advantage of low maintainability, and high scalability & reusability
Easy to adapt and update new policies
Modular architecture for better compatibility to add more complexity
Sungrove Video Documentation 4 Screens.avi
SUNGROVE: Solar Unmanned GROund VEichle
Completely Energy autonomous
Works in urban and remote areas
Maximum solar exposure per footprint
Maximum solar exposure with optimal tilting angle
Optimized hardware design for maximum travel range
Objective Function
States Equations
Co-States Equations
Minimizing Accumulated Heat During Re-Entry of a Spacecraft
This is a well-known but rather difficult problem of minimizing the accumulated heat (eq. 4,5) in a spacecraft during its reentry into the Earth’s atmosphere
This highly nonlinear dynamics problem is govern by three complex equation 1,2,3 of shown in images
This problem has three states and three co-states:
The velocity of the spacecraft V,
The normalized height ξ,
The flight path angle γ,
The accumulated heat
Indirect multiple-shooting method is used, red shows initial values (Iter. 1)and blue line shows converge values (Iter. 78).
Shape Image Processing
Original Image with multiple rudimentary shapes
Histogram analysis of the image
Masted the Image to determine the foreground
Count the all of shapes
Find & locate all the circular shapes
Find % locate all the green shapes
IIMU pitch and roll angles without external vibration
IIMU pitch and roll angles with external vibration
Kalman filter
Complementary filter
Data collection and preprocess
IMU sensor Orientation Estimation Using Kalman Filter
Collect IMU data at 100Hz, preprocess & convert pitch θ and roll ϕ angles in degree
Apply Complementary filter on using gyroscope and acceleration data from the IMU
Use Kalman filter on IMU data for more accurate orientation
For robustness validation, a heavy forced periodic vibration was applied using vibration motor on the sensor
The result are shown in figure with raw data, complementary filter data and Kalman filter data.