Artificial intelligence
Artificial intelligence
Smarter, Thinking Machines to Act Like Humans

For over 30 years, our automated quality control systems and test benches have been making decisions to ensure effective control. Today, technologies are changing and expectations are higher: we no longer want to program machines, we want them to learn on their own. It is Artificial Intelligence.

Everyone talks about Artificial Intelligence, but what is it?

Is it really possible to perform 100% error-free quality control with this technology?

QMTMesure software
Artificial Intelligence
A trendy but often misunderstood subject

Modern technological concepts such as Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL) and Data Science (DS) have become trending topics that everyone talks about but few people really understand the meaning and the differences.

We summarize these concepts here to better explain their potentials. The main interest of artificial intelligence is to transmit human intelligence to machines. It aims to make machines smarter and more intelligent so that they act like humans. Autonomous cars and robots are the best examples of AI.

Machine learning

Machine Learning is a subset of AI that focuses exclusively on predictions based on experiences. It allows the machine to make a data-based decision rather than explicitly scheduling a specific task to run. ML algorithms are purpose-built, learning and improving over time. There are 3 types of ML, namely

  • Machine Learnig with (semi) supervised learning
  • Unsupervised Learnig Machine
  • Machine Learnig by reinforcement
Deep Learning

Deep Learning is a learning approach using neural networks. Artificial neural networks are inspired by neurons in the human brain. They are made up of several artificial neurons connected to each other. The higher the number of neurons, the "deeper" the network. Within the human brain, each neuron receives approximately 100,000 electrical signals from other neurons. Each active neuron can produce an exciting or inhibitory effect on those to which it is connected. Within an artificial network, the principle is the same. Signals travel between neurons. However, instead of an electrical signal, the neural network assigns a certain weight to different neurons. A neuron that receives more load will exert more effect on adjacent neurons. The final layer of neurons emits a response to these signals. Data science is a multidisciplinary term for a collection of tools and techniques for processing data and developing algorithms to solve complex analytical problems. Initially, the goal is to identify hidden trends in the raw data to help improve performance.

The link between Artificial Intelligence and Data Science

Data science is a multidisciplinary term for a collection of tools and techniques for processing data and developing algorithms to solve complex analytical problems. Initially, the goal is to identify hidden trends in the raw data to help improve performance.

It is important to note the need for a large volume of data to effectively implement AI. There is therefore a very close link between AI and Big Data. The use of a DataLake in combination with an AI system is particularly suitable (see DataLake article )



Artificial Intelligence at QMT
A pragmatic and realistic implementation

QMT offers mechatronic multi-inspection (M2IS) solutions and systems for quality control and testing and we are committed to providing the best technologies to meet the customer's need. In line with this strategy, we offer solutions based on Artificial Intelligence and the valuation of data integrated into our solutions.

Our philosophy

The AI integrated by QMT does not replace humans but assists them. To do this, the machine takes care of repetitive, high volume tasks. Humans thus have more time and resources for value-added control and testing. The goal is to promote human work, to minimize waste and to provide elements for the continuous improvement of processes through testing and quality control.

Our approach

When a need of our customer can potentially be met through an AI-based approach, we address the challenge by following the QMT AI checklist. We try to answer the following questions with our client. Is the task:

  • Intensely manual
  • At large volume
  • Structurally repetitive
  • regression or classification
  • With measurable success?

If the answers are positive to these questions, we define the specifications and the means of measuring success together with the client. For the design and implementation of the solution, our experts can be assisted by our partners to offer the best solution.

An example of application: QMT Digital Guardian

Digital Guardian is an autonomous and intelligent device that observes very repetitive tasks, performed by a machine or by an operator. In a first phase, Digital Guardian actively learns the usual cycle of the task. When it has learned enough, it can move on to the monitoring phase where Digital Guardian can detect anomalies and provide information about the sensor and the signal that triggered the alarm. Digital Guardian is in development and we are open to partnership to have our first clients to exploit the possibilities.

This innovation was selected as a finalist for the 2019 CSEM Digital Journey

Partagez cette page
sur votre plateforme favorite