Georgi Tinchev - Personal Website

I'm Georgi

Robotics Enthusiast - Laravel Lover - Software Engineer - Android Apprentice

Portfolio

All
Big Data
Machine Learning
NASA Worldwind
Reinforcement Learning
Robotics
Software Engineering
Vision
Web

Husky – Vision-based motion estimation using a mobile robot

Probabilistic Modelling and Reasoning

Robotics Learning and Sensorimotor Control

Machine Learning & Pattern Recognition

Blog

12th May 2015
Hello world!

Welcome to WordPress. This is your first post. Edit or delete it, then start blogging!

Contact Me

Name

Email Address

Subject

Message

Husky – Vision-based motion estimation using a mobile robot

In this thesis we compare multiple approaches for producing an accurate motion estimate. We provide comprehensive analyses of the performance of an adapted LIDAR Odometry algorithm in comparison with a Visual Odometry system. Previous work fails to perform in-depth analyses between the two techniques. Moreover, we modified an algorithm proposed by Zhang…

Client: The University of Edinburgh
Date: 27th August 2016
Categories: Robotics / Software Engineering / Vision

Learn More

Probabilistic Modelling and Reasoning

Inferring results of upcoming matches of AlphaGo vs Lee Sedol based on previously seen date using probabilistic model. Uses simple factor graph representation to begin with and scales to the TrueSkill model using Expectation Propagation (EP). More information at https://github.com/gtinchev/pmr

Client: The University of Edinburgh
Date: 11th May 2016
Categories: Machine Learning

Learn More

Robotics Learning and Sensorimotor Control

Baxter inverse kinematics simulation in VREP, PCA analysis on targets evince dimensionality reduction from 3D to 1D. Videos available at: http://tinyurl.com/rlsc-question-b and http://tinyurl.com/rlsc-question-d More information on https://github.com/gtinchev/rlsc

Client: The University of Edinburgh
Date: 11th May 2016
Categories: Robotics

Learn More

Machine Learning & Pattern Recognition

Predicting missing pixels using linear regression with adjacent and all pixels, RBF regression with adjacent pixels and neural net with all pixels. Building a model using unreliable training data - fitting a baseline model, labelling noise model, hierarchical model and MCMC. More information at https://github.com/gtinchev/mlpr

Client: The University of Edinburgh
Date: 11th May 2016
Categories: Machine Learning

Learn More

Reinforcement Learning 2

Comparing Q-learning with table lookup for q-values vs features and value function approximation in a 5x5 grid. Extending the problem to 10x10 and 25x25 using RBF functions. More information at https://github.com/gtinchev/rl_2

Client: The University of Edinburgh
Date: 11th May 2016
Categories: Reinforcement Learning / Robotics

Learn More

Reinforcement Learning 1

Comparing SARSA to Q-Learning reinforcement learning algorithms in a 5x5 grid for a Taxi problem. Video preview: More information at https://github.com/gtinchev/rl_1

Client:
Date: 11th May 2016
Categories: Reinforcement Learning / Robotics

Learn More

Advanced Vision 2

Fusing different pictures of a scene taken with RGBD data with different extrinsic camera parameters in order to produce a 3D point cloud as the model of the environment. More information at https://github.com/gtinchev/av_2

Client:
Date: 11th May 2016
Categories: Robotics / Vision

Learn More

Advanced Vision 1

Tracking people dancing ceilidh from a video captured from top in the Informatics Forum. Video preview: More information available on https://github.com/gtinchev/av_1

Client: University of Edinburgh
Date: 11th May 2016
Categories: Robotics / Vision

Learn More

Extreme Computing 2

More advanced MapReduce tasks - tf-idf More information available on https://github.com/gtinchev/exc2

Client: University Project
Date: 2nd April 2016
Categories: Big Data

Learn More

Extreme Computing 1

Basic MapReduce tasks. More information available on https://github.com/gtinchev/exc

Client: University Project
Date: 2nd April 2016
Categories: Big Data

Learn More

Panda

Panda is the differential drive robot, developed for the Robotics: Science and Systems course at the University of Edinburgh. It's an autonomous robot that distinguishes between resources in a maze, captures the correct one and returns back to its base. Project gallery is available at: https://goo.gl/photos/Nm987Lq6Q7zRdaQWA

Client: The University of Edinburgh
Date: 26th November 2015
Categories: Robotics

Learn More

BeeBomBee

BeeBomBee is an artwork purchase/leasing platform, developed in Laravel, using multiple payment gateways (Stripe/GoCardless) and accounting platform (Kashflow) to provide a complete solution. It is designed to ease customers' experience when selecting artwork for their business/personal use. Visit https://beebombee.com

Client: Fiona Innes
Date: 26th November 2015
Categories: Software Engineering / Web

Learn More

Physio-therapy.co.uk

Inverurie Chartered Physiotherapy Clinic is now recognised in Grampian as a centre of excellence. They provide the highest standard of clinical care within a modern facility which includes five treatment rooms, a rehabilitation area and two Pilates studios. The ICPC is a business orientated WordPress website, developed with Ikiji Ltd. Visit…

Client: Inverurie Chartered Physiotherapy Clinic
Date: 4th June 2015
Categories: Web

Learn More

TopCoverage

TopCoverage is a system which allows its users to model the coverage of a mobile network and try out different configurations by altering the signals of currently existing transmitters. The system enables its clients to introduce new transmitters and calculates how a selected area should be populated with antennas. The…

Client: The University of Aberdeen
Date: 23rd May 2015
Categories: NASA Worldwind / Software Engineering / Web

Learn More