Details and application here.
Internationally recognized generalist engineering school of the IMT (Institut Mines-Télécom), leading French engineering school (Technological University), IMT Atlantique aims to support transitions, train responsible engineers, and use scientific excellence to serve teaching, research and innovation.
As part of the Autonomous Pack project led by packaging tracking start-up GoodFloow, in collaboration with researchers at IMT, INRIA and IRCICA, and which aims to develop reusable, traceable industrial packaging that can be shared between manufacturers, we are recruiting a deep learning Post-doctoral researcher in our Mathematical and Electrical Engineering (MEE) Department.
Using these reusable packaging is becoming more and more common in logistics chains. Its impact is beneficial economically, as it reduces costs related to the purchase of disposable packaging. It is also beneficial for the environment, as the emissions associated with the entire life cycle of the packaging are much better than for disposable packaging. However, companies that use these reusable packaging do not have real-time information on the state of the packaging, which can lead to financial losses in case of loss or damage to the packaging. The objective of this project is to develop solutions to monitor the state of these reusable packaging using IoT sensors and deep learning techniques embedded in the sensors.
During the preliminary work, neural network models were developed to perform simple tasks using accelerometer data. They allow to classify some simple events of the packaging life cycle. However, these models are not robust enough to be used in a real environment. Indeed, accelerometer data is not sufficient to obtain accurate classification. It is therefore necessary to merge accelerometer data with other sensors to obtain better results.
While there is literature for similar problems, there is no solution that is directly applicable to our problem. It is therefore necessary to carry out a synthesis of existing work and to adapt it to the Goodfloow context. Under the functional responsibility of the project managers, and in close collaboration with the teams involved in the project, you must identify additional relevant sensors to collect more complete information on the package environment and implement this fusion in deep learning models. One of the research themes will also be to work on the embeddability of these models, as well as their robustness.
Besides, you will participate in the valorization of the results of this work, by producing scientific publications for international journals and conferences.
You have skills in Python, neural networks and PyTorch.
You have a good knowledge of linear algebra and statistics.
You have good listening, analysis and synthesis skills, and are curious and open-minded.
You are adaptable, autonomous, rigorous and able to manage your priorities.
We are looking for people with a PhD in machine learning, deep learning, data science, computer science, obtained less than 3 years before the date of hire, with a strong interest in machine learning.
Deadline for application : 31/05/2024
Start of the contract: As soon as possible
Interviews : real-time
(c) GdR IASIS - CNRS - 2024.