Abstract:
This includes various assignments and project work associated with the Course Work taken at Northeastern University named as Robotics Systems and Science under Prof Marsette Vona. It encompassed various areas ranging from basic linear and angular motion to complex obstacle avoidance algorithms like bug - 1 and bug - 2 and to the final project which made use of APRIL TAGS for localisation. Finally, Rock, Paper, Scissor game was embedded for fun Human - Computer Interaction.
The project URL describes all of these in detail and also includes videos of the implementation
Period:
January 2014 - May 2014
Guidance:
Under the Guidance of Dr. Marsette Vona
At Northeastern University
Project URL:
https://sites.google.com/site/abhishek16990/home/projects/implementing-robotics
Employing Low-Cost Sensors for High Precision tasks using HPC
Abstract:
The project aimed towards using High Performance computing with Low-Cost sensors in order to reduce the overall cost of any project, specially those employing heavy amount of sensors.
RANSAC Line Fitting Algorithm was chosen so as to display results. High Performance Computing was used, in which various parallel computing techniques were incorporated to obtain high precision and speed through these cheap sensors
OpenMP, MPI, CUDA, Matlab PCT and Compiler Switches were used to optimize and accelerate the algorithm. Project was concluded with flying colors, where upto 10 times speedup was obtained by OpenMP, MPI, PCT and upto 400 times speedup was obtained using CUDA
Period:
January 2014 - May 2014
Guidance:
Under the Guidance of Dr. Miriam Leeser
At Northeastern University
Project URL:
https://sites.google.com/site/abhishek16990/home/projects/ransac_hpc
Restricting Growth of Smart World, Memory Systems – Is On-Chip Memory a solution?
Abstract:
Incorporated an extra feature in "DineroIV" Cache Simulator to simulate Scratchpad memory i.e. an extra local storage memory space incorporated in modern Embedded Systems.
The project firstly analysed the bottle neck of today's smart world computer model i.e. our traditional Memory Systems and then tries to put forward some of the latest technologies to cope up. In particular Scratchpad technique was studied in detail.
Lastly, DineroIV was extended to simulate Scratchpad memory as well, and then finally its performance was evaluated on the basis of three benchmarks in thirteen configurations
Period:
September 2013 - December 2013
Guidance:
Under the guidance of Dr. David Kaeli
At Northeastern University
Project URL:
https://sites.google.com/site/abhishek16990/home/projects/on_chip_memory
Abstract:
The project started with a small survey of many taxi and personal drivers, with a question that how many signal/sign boards in a highway are observable and weather in a speed of 100-150 km/h it is noticeable or not. The result was astonishing where only 3% said that it is observable.
The solution to this problem that clicked was to have a speaker that can read out the signal inside a car. The reader here used was a barcode reader with output as a voice to the driver. Barcode strip was stuck in the highway road near the signal carrying information of the signal board. A micro-controller uses this barcode to read out the specified signal using a pre-stored memory. The project aims at the minimum expenditure of the government of just adding a barcode strip to signal, and the device can be made available to all the frequent highway traveler.
Period:
August 2013 - May 2013
Guidance:
Under the Guidance of Dr. Sunanda Manke
At Barkatullah University, Bhopal, India
Group Members : Sachin Jaimini, Priyanshu Gupta, Arun Gannore