Robotics Meets Inclusion: The Technology Behind VIRAS

VIRAS (Visually Impaired Robot-Assisted Shopping) is an autonomously navigating shopping cart that uses AI-powered object recognition, 3D mapping, and barrier-free navigation to safely guide users through the supermarket and independently identify products.

System Architecture

At the heart of VIRAS lies a modified shopping cart system optimized for autonomous assistance functions. It features a modular platform that centralizes all control, navigation, and interaction functions. Movement is motorized, but the system only responds to active user input. This safety-critical principle ensures that the cart moves only when the user explicitly initiates it – an essential feature for trust and control when using the system.

To determine its spatial position, VIRAS uses a localization and mapping process based on stereoscopic depth perception. The environment is continuously captured as point clouds and rendered into a dynamic 3D map. This map is updated in real time, allowing the cart to navigate the store precisely without relying on external infrastructure. The system is designed to operate reliably in highly dynamic environments such as supermarkets, taking into account changing light conditions, variable customer traffic, and temporary obstacles.

The system is complemented by an accessible smartphone app that serves as the central user interface. Users can enter their shopping list, control the cart, and manage other interactions through the app.

Validation and Further Development

VIRAS is currently being tested under real-world conditions in a Migros store in Hinwil. The goal is to evaluate and further develop the system in a practical setting. Key performance indicators such as navigation accuracy, product recognition rate, system latency, and user acceptance are continuously tracked and analyzed.

Thanks to its modular architecture, VIRAS can be flexibly scaled to additional store locations and adapted for new user groups, including individuals with limited mobility or older adults. In the future, the system could also be deployed in more complex infrastructural environments.

For unpackaged goods such as fruits and vegetables, the shopping cart is equipped with an integrated scale. The weight is recorded automatically, and the price is communicated through the app. The entire shopping process can optionally conclude with a self-checkout. Products inside the cart will be automatically recognized by the cameras, matched with the shopping list, and digitally processed for payment.

In the future, the system will include dynamic inventory management. By linking product recognition with a digital twin, real-time product availability can be detected and fed back into the system. Location-based promotions and personalized product recommendations are also planned, contributing to an inclusive, efficient, and intelligent shopping experience.

Go to VIRAS Overview