Ten-Year-Old ‘Pepper’ Robot Reimagined as AI-Powered Tourism Guide by German Engineering Students

Bremerhaven students use Gemini AI to transform a 4-foot-tall Pepper robot into an interactive tourism guide capable of reading gestures and guiding visitors.

By: AXL Media

Published: Mar 28, 2026, 1:21 PM EDT

Source: Information for this report was sourced from Bojan Stojkovski, AI and Robotics News

Breathing New Life Into Legacy Robotics

The "PepperMINT" project, led by computer science and business informatics students in the German port city of Bremerhaven, represents a significant technical bridge between aging hardware and cutting-edge artificial intelligence. The team took a 10-year-old version of the Pepper robot—a 4-foot-tall humanoid originally designed by SoftBank—and updated its antiquated operating system to handle modern software demands. This integration required the students to bypass a decade of software drift, proving that older robotic platforms can still provide immense value when paired with contemporary large language models (LLMs).

AI Personality and Behavioral Guardrails

To ensure the robot remains a focused tourism expert, the students utilized detailed system prompts to define its personality and knowledge base. According to student developer Qusay Mohammad Alhasan, Pepper is programmed to be friendly but firm, strictly limiting its conversation to topics regarding Bremerhaven, the university, and local events. If a visitor asks a question outside these parameters, the robot politely redirects them to its primary function. This level of control is facilitated by the integration of Gemini for speech processing, though the team noted that locally hosted models could also be utilized for increased data privacy.

A Multi-Modal Interaction Experience

Pepper’s utility as a guide extends beyond simple voice responses. The robot is equipped with a chest-mounted tablet that displays visual information, such as city maps and QR codes for preplanned tourist routes. To make interactions feel more organic, the students programmed a library of pre-set gestures, including waving and pointing, which trigger in sync with its vocal responses. Furthermore, the robot’s sensors allow it to read basic human gestures and facial expressions, helping it gauge a visitor's engagement level or potential confusion during a check-in process at the cruise terminal.