6-11-2023 (BEIJING) A team of Chinese scientists has achieved a significant milestone in drone technology by developing drones that can engage in “group chats” and collaborate with each other, emulating human teamwork. This breakthrough has vast implications for enhancing security patrols, disaster rescue operations, and aerial logistics, according to the researchers.
While existing communication strategies for drone swarms often mimic the behavior of bee and ant colonies, the Chinese team has taken a different approach by enabling drones to communicate and collaborate in a manner similar to humans. This innovation not only improves the transparency of the machines’ decision-making process but also enables researchers to gain a better understanding of their behavior.
The pioneering work was carried out by Li Xuelong and his team at the School of Artificial Intelligence, Optics, and Electronics at Northwestern Polytechnical University in Shaanxi province. By integrating large language models like ChatGPT into practical applications, Li and his team have brought these models “to life,” as stated in a WeChat post from the university’s official account.
The post included a demonstration video showcasing the capabilities of a team of five drones. In the video, the drones successfully located a set of keys in an outdoor park. The report emphasized that the drones exhibited key abilities such as human-like dialogue interaction, proactive environmental awareness, and autonomous entity control. The latter refers to the drone cluster’s ability to adjust its flight status in real-time based on environmental feedback.
Each drone is equipped with a “human brain,” enabling them to engage in natural language conversations with one another. The team achieved this capability by leveraging an open-source large language model called InternLM, originating from China.
The dialogue capability allows effective communication between operators and drones, breaking down communication barriers between humans and machines. In the key-finding experiment, three drones promptly volunteered their search abilities when tasked by a user, while the other two drones, equipped with grippers, offered to retrieve the keys. The division of tasks was autonomously decided by the drone cluster.
Once the keys were found, the drones shared images with the user via the group chat for confirmation. This high level of dialogue during critical moments significantly enhances stability and safety when executing complex tasks, as highlighted in the report.
Equipped with multiple sensors, algorithms for low-altitude search, dynamic obstacle avoidance, and visual positioning, these drones perceive their surroundings from various angles and positions. This capability enables them to efficiently collect data and carry out tasks. Termed “proactive environmental awareness,” this feature allows the drones to understand and adapt to their surroundings.
During the key-finding experiment, each of the four drones was assigned a specific area to search. They coordinated their movements to cover these areas efficiently, generating a simplified map of the terrain to guide their efforts. Importantly, the drones were also able to identify and avoid human operators in their flight path, ensuring safer operations.
The report highlighted the potential applications of this technology in security inspections, disaster relief operations, and drone-based transport and logistics.
Previously, Li’s team explored optics-driven drones that utilize high-energy lasers for remote power supply, offering them potentially limitless endurance. In October, Li spearheaded the development of a drone guidance system for underwater operations, named Navigator.
