Week 5 (18th March 2019/ 22th March 2019)

With the arrival of infrared sensors, it was necessary to build a PCB to connect them to the Arduino and a holder to fix them on the top of the robot. An ADC converter was required to have more analog inputs on the Arduino. For this we bought the ADS1015 which has 4 analog inputs that will be used for the 4 sensors and that connects to the Arduino through I2C. The PCB was also used for ADS1015 and ESP-01. The sensors support was made of nylon being a parallelepiped of dimensions (14mm * 14mm * 60mm) that will accent on the top of the PCB.

I have advanced some functions of the learning code so that I can start testing when the robot is fully ready.

Advertisements

Week 4 (11th March 2019/ 15th March 2019)

This week the first meeting was held with the supervisor and the co-supervisor and a demonstration was made using the ultrasound sensor so that the robot could navigate without encountering obstacles.
Infrared sensors, Sharp GP2Y0A41SK0F, were purchased with a range of 4 to 30 cm and it was thought possible to create a network with a Router to use the university network.
Some points of the thesis were advanced while the sensors did not arrive.

Week 3 (4th March 2019/ 8th March 2019)

During this week I created two ROS nodes in which one receives the information from the ultrasound sensor by posting it on a topic and then sends movement commands to the robot. The other node subscribes to the topic where the sensor values ​​are stored and prints those values ​​to the terminal. This program is an initiation on my part to both ROS and WiFi communication and is the basis of what will be done later. The second node will be responsible for learning while the first node is only a link between robot and computer.
Several possibilities of actions arrays have been thought out, but nothing yet in concrete. For now the robot has 8 actions (walking forward, walking backward, turning 45º, 90º and 135º left and right), which does not appear to be 100 % effective.
On March 8, LAR Metting was held where, together with my colleagues, the dissertations were presented, explaining the objective and motivation of each one, as well as the plan drawn up to conclude it.

Week 2 (25th February 2019/ 1st March 2019)

During this week the new WiFi module arrived and the communication between PC <-> Robot was established using the AT commands that come incorporated in the ESP-01.
The sketch of the maze was also drawn as well as all the planning of dimensions and material to be used. The complete maze will have a size approximately 2m * 2m if it is decided that we will go this way. The walls will have a thickness of 2cm and a height between 12 to 15cm and will be made of wood so that it is possible to have a reflection in the same direction in the case of the infrared sensors that will be bought.

Week 1 (18th February 2019/ 22th February 2019)

During this week I finished the ROS tutorials that were requested during the weekend, I gave my supervisor the 2nd version of the intermediate presentation and finished the intermediate report. Due to the problem of WiFi communication I’ve been researching more about how to solve the problem. I did more intensive research on the model-based, especially the Dyna-Q and Prioritized Sweeping algorithms as well as a research on different methods of RL and transition models. On the 21st and 22nd of FEBRUARY I participated in the first two sessions of the ROS workshop, given by Professor Miguel Oliveira.