1 Introduction

In societies rooted in sight, vision is the primary sense and plays a vital role in all aspects of life. However, the World Health Organization reports that a total of 253 million people worldwide suffer from visually impaired diseases, of which about 36 million are blind.

Most ordinary people's desire for AI is that it can help humans share the workload while providing more comfortable entertainment when they are relaxing, and more advanced, ambitious scientists, hope that AI can revolutionize the world, which is a qualitative leap. However, the blind community longs for AI to lead them to better integration into the world.

Nowadays, there are already AI technologies that can bring some warmth to the blind, such as voice assistants, which can help the blind carry out various operations in their daily lives, such as sending text messages, making phone calls, setting reminders, and so on. Visual aids can help the blind recognize objects, faces, and words through image recognition and voice conversion, enabling the blind to understand their surroundings. The navigation system helps them find their destination in unfamiliar environments through voice prompts and vibration feedback. Reading aids that convert textual content such as books and articles into speech or Braille allow the blind to access information through auditory or tactile means and enjoy the pleasure of reading. These AI applications can help blind people better integrate into society and improve their quality of life and autonomy. Of course, this alone is not enough. Most blind people are confined by their feet and find it difficult to take that step.

Guide dogs, as the eyes of the visually impaired, can serve the visually impaired and guide them on their trips, but the number of blind people is very different from the number of guide dogs. For example, the number of existing guide dogs in China is less than the number of pandas, a national treasure, and the reason for this is that guide dogs require rigorous breeding and training of breeds, and the training of a licensed guide dog is far more difficult than we can imagine. They have to learn to go up and down steps, avoid obstacles, cross the street to identify the owner's usual places and surroundings, and other skills, but also have a smart, obedient, and friendly character as long as the saddle on, you can immediately enter the work state, become the eyes of the blind. These factors profoundly explain the fact that there are only a handful of guide dogs that can provide services to the visually impaired.

In recent decades, with the rapid development of artificial intelligence and automation technology, to alleviate the contradiction between the large number of visually impaired people and the scarcity of guide dogs, the Guide Dog Robot (GDR), a robot that possesses important features of a guide dog, has been created, aiming to replace or even surpass the function of a guide dog in many aspects. The GDR is a robot with important features of a guide dog that combines sensor technology, intelligent algorithms, human–computer interaction, or other advanced technologies to provide safe path guidance for the visually impaired. As an alternative to a guide dog, a GDR can naturally realize the key features of a guide dog. So far, most of the features unique to a guide dog have been realized by the GDR, which is the mainstream direction for the development of assistive devices for the blind due to the excellent expansion potential of GDRs' features and their ability to reduce the probability of injuries to the blind. In addition to this, GDR can even perform tasks that traditional guide dogs cannot, such as navigating to new and unfamiliar places. The sensors can detect the surrounding road conditions, process the sensor information, and then send the corresponding commands to the GDR. Once the sensors capture the surrounding information, combined with the path planning algorithm, it will choose the optimal route to the destination.

From the above, it can be seen that the existing GDR can indeed fulfill the main tasks of a guide dog, and even have functions that a guide dog does not have. To a certain extent, the maturity of GDR is recognized. However, functional completeness cannot be directly equated with the fact that GDR will be accepted by the visually impaired. Undoubtedly, no blind person would like to be guided by a rigid robot. The performance of traditional guide dogs in terms of dynamic obstacle avoidance, coordinated guidance, and technical reliability is more likely to be convincing. I am a guide dog robot developer and I am also highly nearsighted. When we think about it differently, this concern is completely understandable. Visually impaired people lack perception of the outside world, so they have no fear of handing over their safety to a robot, by the robot's guidelines for all activities, require not only trust and courage but also a variety of guarantees.

1.1 Guarantee 1: technical safety and reliability

Technology is the first factor for GDR to be accepted by the blind as early as possible. The commercialization of GDR should be supported by a wide range of issues which include battery, overheating, motor strength to avoid hurting the user, reliability, repair, life cycle, disposal and environmental costs. The whole stage of GDR development and testing is no less than the stage that a car goes through from R&D to market and then to be purchased and driven by users. Moreover, the development process of automobiles is already mature, while GDR is still in a state of crossing the river by feeling the stones. With technical reliability, GDR has more bottom line. In the future, mobile GDR will become a part of urban intelligent transportation. Based on the vehicle networking system, mobile GDR can exchange information with moving vehicles.

1.2 Guarantee 2: emotional companionship and support

Long-term companionship requires emotional support. Humans get emotionally attached to all sorts of objects, from fountain pens to dolls, why not machines? Some popular technologies will help GDR to enrich the collection of user information and shape a more intelligent service robot. In addition to accompanying and guiding them at their side, they can also monitor the user's travel information, health status, reminders of travel conditions, weather, etc. in real-time, facilitating personalized services. As the blind person's "cooperation" with the GDR deepens, it helps to increase human–robot trust, which in turn facilitates and enriches positive emotional communication between humans and robots.

1.3 Guarantee 3: collaboration and assistance on the team

When the user's unexpected accidents, such as falling, being hit, etc., the GDR can take measures to timely alarm, seek help from the surrounding environment, and promptly inform the family of the current situation. At the same time, to prevent the GDR from program errors and failures that may arise during the actual work, there must be live human staff monitoring the GDR via a cloud backend support platform. Once the GDR malfunctions, the person should be the first to contact the visually impaired person to stay in place and wait for help. Artificial intelligence is not completely detached from human beings, and in the future, the Human-GDR collaboration model will play an even more critical role in guaranteeing the safety of the visually impaired.

1.4 Guarantee 4: legal system and rules

Visually impaired people, especially the blind, usually do not get enough attention from the outside world. However, no small group of people should be abandoned. At the right time, the law should be supplemented to protect the legal rights of the visually impaired and GDR. As with self-driving cars, there are ethical issues at the engineering level that are difficult to cope with, such as the division of responsibility after a GDR navigation accident. Even a simple robot can raise serious legal and ethical questions. Who should be held liable for damage caused by improper maneuvers involving robots?

The current legal regulations for GDR in various countries are not mature, and this is something that requires further thought and discussion by governments. Some directions to solve the problem could be that with the development of 5G technology, the GDR data will be transmitted to the cloud via 5G technology to save with the implementation of the guardianship, and the video recordings during the work will be backed up. In addition to cloud backup you can also store data offline, such as removable hard disk type of hardware storage, which can alleviate the pressure generated by too much data in the cloud. These can be used as a basis for determining any liability disputes that may arise later. However, GDR based on AI systems collects large amounts of sensitive data during their work, for example through facial and voice recognition functions. There is considerable scope for abuse in this area, not only in terms of personal data protection but also in terms of information, privacy, and dignity. How to balance the guardianship of data in the cloud with the protection of privacy for the visually impaired is an issue that needs to be analyzed in depth. When these issues are discussed with clarity, clear legal provisions are a must.

When I asked the ChatGPT what else the AI could do for the blind in the future? Its answer was surprisingly no different from what existing ones can do for the blind. I was disappointed by the AI's imagination in this regard; did it mean that in the future the AI would still serve the blind in these ways, or would do it better? I hope it is the latter. So the answer to the question of how AI can help the blind in the future remains to be answered. In the future, with the development of AI and the maturity of GDR's functions, it is expected that an equal, non-discriminatory, and friendly world will be built for the blind so that they can take their first steps with confidence under the guidance of GDR.