What are the specifications for the "Goldorak" project?
The project spans a range of different fields including embedded systems and robotics, connected objects and IoT, the factory of the future, digital twins, cybersecurity and AI.
The aim is to design, programme and simulate an architecture representative of a modern production line comprising DOBOT Magician robotic arms. With that in mind, we have had to study how the robotic arms worked, develop programs (using the Python and C languages) for controlling them and create a secure environment. We also set up a simulation environment and control environment remotely (broker MQTT). This enabled us to continue working on the project during lockdown in the fall of 2020.
What skills did you use?
We used the skills developed during our two specialization pathways: "Observation systems and Artificial Intelligence" (SOIA) and "Security and Digital Systems" (SNS): AGILE organization for managing the project, development using the Python and C languages, image processing and supervised learning mainly.
What challenges arose?
One of the challenges was obviously the project’s broad thematic scope, but we managed to organize ourselves to share out the tasks. As such, the main challenge we encountered was having to adapt our working method during lockdown.
For a month and a half, we could no longer access our work room you see, which meant we could no longer work physically with the robotic arms. We therefore had to adjust our goals and develop a simulation of the robotic arms and of our working environment. Despite the extra time we had to spend doing this, it nevertheless turned out to be an undeniable advantage for our project since, in this way, it chimed with the “Hardware-in-the-loop” philosophy: simulating the devised systems before producing them in practice.
How is the teamwork organized?
We are a team of five students: Marie-Amélie Defresne and Adrien Grivey are doing the "SOIA" major, and Mahdi Salameh, Ayman Al Hajjar and Ayrwan Guillermo the "SNS” major.
In keeping with the AGILE method, ahead of each Sprint we establish a Backlog where we indicate all of the tasks to be done during the fortnight of work. We then divide up the work depending on each member’s aspirations and skills.
What’s your verdict of the project?
Given the exceptional circumstances in which we had to work this year, it wasn’t always easy to keep to the Sprint goals, but overall we were able to adapt.
As such, today we’ve achieved most goals: controlling the operation of the robotic arm, having functional control codes and an effective environment detection algorithm. The final weeks of the project were spent finalizing the remaining tasks, not least the link between simulation and reality.
In personal terms, we all agree that the project has taught us a lot about AGILE project management and team organization.
What’s more, as far as the technical gains are concerned, we were able to learn about new working tools like Coppelia Sim, supervised learning, setting up MQTT brokers or using an API to control the DOBOT Magician.