Tokyo (SCCIJ) – The Swiss software boutique consultancy Vigience with offices in Tokyo, Horgen (Switzerland) and San Francisco (USA) has transformed the famous Softbank robot Pepper for the first time globally from a “meet and greet” shop assistant into a useful factory worker at a renowned Toyota supplier in Japan. The achievement was made possible by programming Pepper directly and using an artificial intelligence engine. The cute robot now moves independently on the factory floor, supervises the working conditions in the plant and supports workers when a machine in the production line malfunctions. SCCIJ talked to Vigience founder and CEO Markus Stierli about this progressive project and how it has changed the course of his company.
Pepper usually talks to guests in shops, sells Nescafé coffee machines or greets hotel guests. How did this Japanese robot become a factory worker?
Stierli: The automotive supplier Kojima Press Group, a manufacturer of pressed metal and plastic parts for Toyota, wanted to further automate some production tasks in the factory of its subsidiary Maruwa Electronic & Chemical. Their main reason is the growing labor shortage in Japan and they wanted to build a solution that they can share with small and medium-sized manufactures. As industrial robots are very expensive to buy and maintain for small manufacturers, they came up with the idea of using the much cheaper Pepper. Three companies were commissioned for an implementation test. Vigience has been awarded the contract.
What applications have you been able to develop for Pepper?
Stierli: We looked around the factory and considered what Pepper could do with his own technical capabilities and what functions we could add to make the robot really useful. The first use case we came up with are inspections. Pepper now patrols independently in the factory, measures the temperature, humidity and dust content of the air and saves the data automatically. If there are strong deviations from the average figures, it will sound an alarm. So far this task has been carried out by humans. Pepper himself cannot do very much except welcoming, short conversations, cute arm and head movements and rolling around.
What did you have to change so that the robot could work as an inspector?
Stierli: We use the camera so that Pepper can find his way in the factory. Position finding via indoor positioning system would be too expensive for a medium-sized company, though. Therefore, we have programmed Pepper to recognize lines and code signals taped to the ground. In doing so, we needed to write a software that Pepper uses to move around while making sure it does not overheat any of its motors.
The second project from Vigience for Kojima Press sounds more challenging, as Pepper also has to directly support the human workers. What application did you develop there?
Stierli: The second use case is based on artificial intelligence. When a machine or an industrial robot in the production line stops working, it sends a visual alarm to Pepper. The robot rolls automatically to the worker at the machine. Such workers are normally not engineers and therefore need some professional advice. The worker describes the technical problems of the broken machine by talking to Pepper and answering multiple choice queries on Pepper’s touch panel (see video).
How did you make this possible?
Stierli: Pepper analyzes the information using the artificial intelligence engine IBM Watson. Previously, we had fed Watson’s AI engine with all possible scenarios of machine damage and the corresponding solutions. In the course of time, the AI engine has become better and better at finding the malfunction reasons and proposing the right solution.
The keyword “artificial intelligence” has become omnipresent. How useful are such machine learning programs in your opinion really?
Stierli: In fact, the current hype is very big. Many people now think, all problems can be solved with artificial intelligence. But AI does not fit everywhere and does not always provide the best answer. Regular mathematics and a simple set of fixed rules are often enough. The actual machine learning algorithm is understood only by very few specialists. But at Vigience, we make it our goal to know when and how to apply AI and how to train the learning algorithm best with data. Our company will now go further into this direction.
Where do you see applications for AI outside of the production realm?
Stierli: The current challenge in the IT sector is the integration of artificial intelligence applications into existing business software (CRM/ERP) solutions such as Salesforce and SAP. As a consultant, Vigience is very well positioned here because we come from this area. Our proprietary “Overcast” application is a simple to install connection between Salesforce and SAP and can pull together the necessary data for an AI engine.
Can you give an example of such an application in the area of business software?
Stierli: For a customer in Switzerland, we are currently working on a project to help this manufacturer with a prediction of future order volume six months ahead. For this, we evaluate the data of the past and let the AI engine learn with this information. The system gradually gains experience like a human being, but much faster because it processes much more data. This allows the AI to do more and also reliable work.
What Vigience does
Salesforce & SAP consulting: Vigience implements digital business solutions on industry-leading products such as from Salesforce, SAP, Microsoft and Vigience, and also customizes and develops new features on these platforms.
Cloud integration consulting: Vigience not only provides end-to-end integration consulting for SAP, Salesforce & other products, but also finished integration apps.
Application Development: Vigience helps companies to build an app for employees or customers or to launch a new SaaS service for their customers from scratch. It provides full stack development services on a range of platforms.
Product Concepts, Designs and Plans: Vigience helps you to envision, architect, design, build and run mobile and SaaS products. It has launched numerous apps, cloud products and add-ons for customers such as NEC, SAP or Vodafone.
Interview: Martin Fritz for SCCIJ; Photo: Kojima Press