XI'AN, March 29 (Xinhua) -- Users can now command a robot dog using nothing but their mental intent: with just a thought, the machine autonomously plans its path, avoids obstacles and navigates precisely to a designated location.

Such a sci-fi scene has now become a reality at Xi'an Jiaotong University in northwest China, thanks to a breakthrough by Professor Xu Guanghua and his team, who successfully integrated electroencephalogram (EEG)-based control with autonomous navigation.

At the heart of the achievement lies the non-invasive brain-computer interface (BCI) technology, which captures electrical signals from neuronal activities to enable precise control of mechanical devices, Xu explained.

He described the system as a kind of "remote control in your mind." When a user forms an intention, such as "move forward," the brain generates corresponding EEG signals. The system collects and decodes those signals, identifies the intended command, translates it into a control instruction, and sends it to the robot dog, which then executes the movement.

Currently, the system supports 11 basic mental commands, including forward, backward and turning, with the potential to expand further. Its recognition accuracy exceeds 95 percent, and the lag between thought and action is only about one second.

Amid a global surge in BCI research, the invasive technologies offer high precision but rely on surgical implantation, carrying risks of trauma, infection, immune rejection, and signal degradation over time -- factors that make them costly and difficult to scale.

In contrast, the non-invasive approach chosen by Xu's team is safe, cost-effective, user-friendly, and well-suited for a wide range of applications, particularly in rehabilitation medicine and consumer settings.

Non-invasive signals, however, are inherently less precise, making continuous, fine-grained real-time control a challenge. Xu noted that requiring users to manually control every movement and posture adjustment would not only be extremely difficult but would also place them under intense mental strain -- defeating the very purpose of technological empowerment.

To address this problem, the team moved beyond the narrow focus on signal precision and adopted a human-machine collaboration model with clearly defined roles. "Humans are responsible only for issuing high-level intentions such as 'where to go' -- the decision-making part that the brain handles most easily," Xu said.

"Meanwhile, high-precision, high-speed, repetitive tasks such as autonomous navigation, environmental perception, dynamic obstacle avoidance, and motion execution are handled entirely by the machine's own intelligent systems," he said.

This approach significantly improves efficiency and system stability, circumvents the limitations of non-invasive signal precision, and maximizes the complementary strengths of human decision-making and machine execution -- bringing the BCI technology closer to practical application.

Xu emphasized that advancing the BCI technology requires both sustained breakthroughs in core technologies and deep integration with cutting-edge fields such as artificial intelligence, autonomous navigation, and intelligent perception. His team's work embodies this dual path: using practical innovation to address shortcomings in non-invasive interfaces while grounding development in real-world needs.

Xu envisions brain-computer interaction systems that seamlessly combine human decision-making with machine intelligence, ultimately making robots capable assistants in daily life.

The robot dog holds promise as an aid for individuals with disabilities, as well as for applications in elderly care, medical assistance, rehabilitation training and intelligent companionship, he said. ■