AGV performance: a deep dive into neural networks – News Couple
LOGISTICS

AGV performance: a deep dive into neural networks


Machine learning, neural networks – big terms in modern logistics, but what do they actually mean? What we need is an expert to tell us. Paul Hamblin, editor of Logistics Business magazine, meets one.

Specialist in AGV and mobile robotics Colmorgan Measures the performance and behavior of AGV systems, both at the system level and on board individual AGVs. He collects information on drivetrains, laser scanners, localization, traffic, and obstacle interference, thus finding trends and patterns that help him improve products and system configurations.

Measure the ups and downs of productivity or daily usage trends, use data to improve routings, increase productivity, and calculate smarter use of resources. For the end user, this translates to lower costs and higher revenue. The collected data can also provide important clues to external processes that may disturb the performance of the AGV. For example, pedestrians getting in the way of AGVs, or manual forklifts driving in areas originally planned for AGVs.

But how do you do all this? Samuel Alexanderson, Director of Product Management for AGVs at Kollmorgen is our patient guide.

“An artificial neural network is a computational model that is loosely based on the structure of the human brain,” he began. “Brain cells, or neurons, are connected to a complex network of nerves through which electrochemical signals are transmitted. Simply put, we can say that if the weight of the input signals is strong enough, the neuron will fire, and the signal will continue to the next group of cells to which the cell is connected. nervousness;

“This way, the structure of connections between all the different neurons in our brain will determine how the signal spreads, and when we learn new things, what actually happens inside our brain, is the restructuring of the connections.

“In an artificial neural network, the signals are digital rather than electrochemical, and communication strengths are stored in weights. Initially, these weights will have a random value, which means that when we feed an input to the network, it will just output random nonsense. But in the same way that By which we humans learn from experience, we can allow the model to learn from experience in the form of data.

Each data point will be an expected input and output, so these pairs are examples to learn from. Using so-called machine learning algorithms to adjust the weights incrementally, will bring the output from the model closer and closer to the expected output, so that it can learn to make predictions.

“For example, if we want to teach a machine to tell the difference between a cat and a dog, we need to create a dataset with images (the inputs), where each image has a label indicating whether it is a cat or a dog (the correct output). After training the neural network to In the data set, we might feed it a new image, and although this image has not been seen by the network before, it will tell us whether it is a cat or a dog.”

Can he provide examples of how ideas are generated from data in a logistical context?

He answers: “Designing an AGV system can be a complex task that requires a lot of skill and experience.” “When designing a road network for example, there can be tens of thousands of individual road extensions that need to be configured correctly. Of course, it is easy to make a mistake and often such errors are not detected until you run the system in a simulation.

“Therefore, we are currently developing tools that can analyze the road network directly, so the user gets more immediate feedback. In AB internal tests, we saw a 5-fold improvement in the time it took for users to find the root cause in a misconfiguration.

“There is a more anecdotal indication of the impact we can have with a data-driven approach, which is a support case where something the engineer has been debugging for hours using traditional methods has been resolved within 10 minutes using a prototype that analyzes the configuration data automatically.”

So we have – as always, it’s all about cats and dogs.



Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button