Lightweight wearable auxiliary blind guiding glasses

Abstract

The application discloses a kind of light weight wearable auxiliary blind guide glasses, belong to visual impairment auxiliary equipment technical field.This equipment includes light weight glasses main body, machine vision identification module, intelligent operation processing module, audio prompt module, positioning navigation module and low-power supply module, the weight of complete machine is controlled within 50 gram, wear no burden.Equipment relies on YOLO-v8 algorithm and light weight big model, can identify the static obstacle in 0.5 meter to 10 meter range, dynamic obstacle and pre-judge risk in real time, with high-precision text recognition technology, realize whole scene text accurate identification broadcast;Satellite positioning and inertial navigation technology are fused, complete indoor and outdoor continuous navigation and small article search;Open type bone conduction sound structure and hierarchical priority voice broadcast system are used, and use safety and information transmission efficiency are considered;All data operation is completed locally, privacy is high, and long-acting endurance of more than 8 hours can be realized by low-power design.The application solves the problems of traditional blind guide equipment, such as heavy, single function, low recognition accuracy, poor privacy, insufficient endurance, high integration degree, strong scene adaptability, comfortable and safe to wear, can be widely applied to visual impairment person daily travel, home life auxiliary scene, has very high practical value and popularization prospect.

Description

Technical Field

[0001] This invention patent belongs to the field of assistive device technology for the visually impaired, specifically relating to a lightweight wearable assistive guide glasses, which is suitable for assistive scenarios such as daily travel, object recognition, scene navigation and environmental safety warning for visually impaired people. It can effectively make up for the functional defects of traditional guide devices and improve the safety and convenience of independent travel for visually impaired people. Background Technology

[0002] According to publicly available data from the World Health Organization and the China Disabled Persons' Federation, the global population of visually impaired individuals is enormous, with over ten million in my country alone. The vast majority of these individuals face difficulties such as independent travel, insufficient environmental awareness, and limited access to daily information. Basic activities for visually impaired individuals, such as daily travel, identification of everyday objects, and reading public information, rely heavily on guide dogs, canes, or assistance from others, severely restricting their independent living and activity range.

[0003] Currently, mainstream guide assist devices on the market have many technical defects and pain points in use: traditional white canes can only detect obstacles at close range underfoot and cannot identify obstacles at height, dynamically moving targets and road markings, with limited warning range and extremely low level of intelligence; guide dogs have long training cycles, high maintenance costs and are difficult to popularize, failing to meet the universal needs of the vast visually impaired population; existing intelligent guide devices are mostly bulky, burdensome to wear, have limited functions, low recognition accuracy and poor privacy.

[0004] Existing smart guide glasses for the visually impaired generally suffer from excessive weight and excessive pressure when worn, easily causing head fatigue with prolonged use. Furthermore, most devices rely on cloud data transmission for recognition and processing, resulting in high response latency and privacy risks such as leakage of user travel data and environmental imagery. Functionally, traditional guide glasses cannot simultaneously handle core functions such as dynamic obstacle trajectory prediction, high-precision multi-scene text recognition, integrated indoor and outdoor navigation, and small item location. Moreover, voice prompts lack priority differentiation, easily leading to information overload and the obscuring of critical safety information, making them unsuitable for diverse usage scenarios such as complex urban roads and indoor homes.

[0005] Furthermore, existing guide glasses for the visually impaired mostly use a closed-back earbud structure for audio output, which blocks out natural ambient sounds, preventing users from perceiving surrounding voices, vehicle horns, and other emergency sounds, posing a significant safety hazard. Simultaneously, the devices have poor power consumption control, with battery life generally less than 4 hours, failing to meet users' daily travel needs and severely hindering the practical promotion and application of intelligent guide glasses. Based on these shortcomings of existing technologies, there is an urgent need to develop a lightweight, highly accurate, low-power, highly private, and functionally integrated wearable assistive guide glasses for the visually impaired. Summary of the Invention

[0006] Purpose of the invention

[0007] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a lightweight wearable assistive guide glasses for the blind. This invention solves the technical problems of traditional guide devices, such as large weight, limited functionality, low recognition accuracy, short battery life, poor privacy, cluttered voice prompts, and limited environmental perception. It achieves multi-functional integration, including accurate warning of static and dynamic obstacles, high-precision full-scene text recognition, integrated indoor and outdoor navigation, small item search, and tiered voice prompts, while also taking into account wearing comfort, safety, and data privacy.

[0008] Technical solution

[0009] To achieve the above objectives, this invention provides a lightweight wearable assistive guide glasses, including a glasses body, a machine vision recognition module, an intelligent computing processing module, an audio prompt module, a positioning and navigation module, and a low-power power supply module.

[0010] The main body of the glasses is a lightweight glasses-style terminal structure, made of high-strength lightweight engineering plastic. The main unit of the device is integrated inside the temples and frame, and the overall weight is strictly controlled to within 50g. There are no bulky external devices, so there is no burden on the head when wearing them, making them suitable for long-term wear.

[0011] The machine vision recognition module includes two sets of low-power high-definition cameras and an AI vision processing unit. The cameras are embedded on both sides of the front of the glasses frame, with no protruding exposed structure, balancing aesthetics and field of view, and can achieve full-coverage image acquisition within a range of 0.5 meters to 10 meters in front. The AI ​​vision processing unit has built-in lightweight YOLOv8 algorithm and lightweight large model, which can perform real-time image analysis locally to achieve multi-dimensional target recognition. The recognized objects include static obstacles, dynamic targets, and key scene information. Static obstacles include steps, roadblocks, walls, stationary vehicles, etc.; dynamic targets include pedestrians, bicycles, and moving cars, and can intelligently predict the movement trajectory of targets and identify collision risks in advance; key scene information includes traffic lights, pedestrian crossing signs, door signs, facial information, etc.

[0012] The intelligent computing module incorporates an edge computing unit, enabling all visual recognition, data processing, and information analysis to be completed locally without relying on cloud servers for data uploads. This fundamentally protects users' privacy data, including travel images and location information. The module integrates a high-precision OCR recognition algorithm, adaptable to complex lighting environments such as strong light, low light, and backlight. It accurately recognizes various text information, including printed and handwritten text, achieving a text recognition accuracy rate of over 97%. It can quickly recognize everyday text information such as road signs, menus, house numbers, and shop signs, and simultaneously convert text to speech.

[0013] The audio prompt module uses an open-back bone conduction speaker, embedded inside the temple of the glasses, fitting snugly against the user's ear. It does not require insertion into the ear canal, does not block out ambient sounds, and completely avoids the safety hazards associated with traditional in-ear audio devices. The module also features a built-in hierarchical priority voice prompt system that intelligently categorizes and sorts all recognized information, prioritizing critical information affecting travel safety such as vehicles, red lights, steps, and sudden dynamic obstacles, while secondary information includes text, shop information, and item locations. The voice prompts employ a progressive playback logic, first providing a risk warning, then detailing the scenario and distance. It also supports user-customizable prompt volume and playback frequency to adapt to different users' auditory habits and usage needs, avoiding information overload that could interfere with user judgment.

[0014] The positioning and navigation module integrates a lightweight GPS positioning unit and an inertial navigation unit, fusing machine vision recognition data with satellite positioning and inertial navigation data to achieve continuous, uninterrupted navigation both indoors and outdoors. The module supports voice interaction, responding to user voice questions and accurately retrieving the locations of nearby amenities. It also features precise indoor item location capabilities, assisting users in locating everyday items such as keys, mobile phones, and wallets. Furthermore, the module supports preset route navigation, providing real-time voice guidance based on the planned route and proactively avoiding obstacles and dangerous areas.

[0015] The low-power power supply module uses an ultra-low power dedicated chip and is equipped with an intelligent energy-saving algorithm, which can effectively reduce the standby and operating power consumption of the device. After a single full charge, it can work continuously for more than 8 hours, meeting the battery life needs of users for daily travel and home use. The power supply module is embedded in the tail of the temple, which is small in size and does not increase the overall weight and size of the device.

[0016] Beneficial effects

[0017] Compared with the prior art, this invention patent has the following significant advantages:

[0018] 1. Lightweight design and excellent wearing experience: The overall weight of this device is controlled within 50g. All core hardware is integrated inside the frame and temples. The structure is compact and there are no external devices. There is no pressure on the head when wearing it for a long time. Compared with traditional smart guide devices, portability and comfort are greatly improved, and it is suitable for daily long-term wear.

[0019] 2. High recognition accuracy and wide scene adaptability: It integrates the advanced YOLOv8 algorithm with a lightweight large model to achieve obstacle recognition in the full range of 0.5-10 meters. It can distinguish between static and dynamic obstacles and predict their movement trajectory to avoid collision risks. With high-precision OCR recognition technology, it is suitable for all lighting scenes and multi-font text recognition, with a recognition accuracy of over 97%, completely solving the pain points of visually impaired people in reading signs, bills, and symbols.

[0020] 3. Comprehensive safety design and thorough travel protection: It adopts an open bone conduction sound structure, which does not isolate ambient sound and retains the user's original environmental perception ability; it is equipped with a hierarchical priority voice broadcast system, which prioritizes the push of key safety information, and the progressive broadcast mode avoids information overload, and intelligently filters invalid information, thereby improving the efficiency and safety of users' travel judgment.

[0021] 4. Indoor and outdoor integrated navigation with high functional integration: It integrates machine vision, GPS positioning and inertial navigation technology to solve the problem of indoor positioning failure of traditional equipment and realize continuous indoor and outdoor navigation; at the same time, it adds voice question and answer, small item search and preset route obstacle avoidance navigation functions, covering the needs of users in multiple scenarios of travel and home life, and its practicality far exceeds that of traditional guide devices.

[0022] 5. Strong privacy and excellent battery life: All core calculations are completed on the local edge computing unit, eliminating the need for cloud data uploads and completely preventing the risk of user privacy data leakage; with low-power chips and intelligent energy-saving algorithms, the battery life can reach more than 8 hours, meeting the user's normal all-day use needs. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0024] Figure 1 : Schematic diagram of the overall structure of this invention patent;

[0025] Figure 2 This invention patent hardware module system architecture block diagram;

[0026] Figure 3 : A schematic diagram of the patent identification and broadcasting process of this invention;

[0027] Figure 4 : Flowchart of the indoor and outdoor navigation function implementation of this invention patent. Detailed Implementation

[0028] The present invention will be further described in detail below with reference to specific embodiments.

[0029] This invention discloses a lightweight wearable assistive guide glasses for the blind. The overall structure adopts a conventional eyeglass design, with the frame and temples made of lightweight, high-strength ABS engineering plastic, making it drop-resistant, wear-resistant, and extremely lightweight. The entire device weighs only 48g, meeting the requirements for lightweight wear. The core hardware includes a high-definition camera, an AI visual processing chip, an edge computing unit, a bone conduction speaker, a GPS positioning module, an inertial navigation module, and a built-in lithium battery. All hardware is embedded and integrated inside the frame and temples, making the device indistinguishable from ordinary smart glasses, highly discreet, and without any sense of being a special device.

[0030] In terms of obstacle recognition and safety warning implementation, the device uses dual front-facing high-definition cameras to collect real-time images of the road environment ahead. These images are then analyzed in real-time using the locally integrated YOLOv8 algorithm and a lightweight large model, with a scanning range covering an area from 0.5 meters to 10 meters ahead. When static obstacles such as steps, roadblocks, or walls are detected, the device broadcasts distance and risk warnings in real time, such as "Steps ahead 5 meters, watch your step" or "Roadblock ahead 3 meters, please detour." When dynamic targets such as pedestrians, bicycles, or motor vehicles are detected, the system automatically calculates the target's speed and trajectory, predicts collision risks, prioritizes emergency warnings for rapidly approaching dangerous targets, and provides routine warnings for slow-moving harmless targets, accurately distinguishing risk levels and avoiding ineffective warnings.

[0031] At the text recognition implementation level, the device's OCR module continuously collects text information from the screen in real time. It can adapt to complex lighting conditions such as strong light, cloudy days, nighttime, and low indoor light. It can quickly recognize and analyze text in various scenarios, including road signs, traffic signs, shop signs, restaurant menus, house numbers, and paper documents. It can reliably recognize both standard printed and handwritten text, with an accuracy rate consistently above 97%. After recognition, it immediately converts the text into clear voice broadcast, helping users quickly obtain environmental text information and solving everyday information reading challenges.

[0032] In terms of navigation and item retrieval, the device integrates GPS satellite positioning and inertial navigation data, combined with machine vision scene understanding capabilities, to achieve accurate navigation on outdoor roads and continuous positioning in indoor no-signal scenarios. Users can initiate commands via voice interaction, such as "find the location of nearby supermarkets" or "navigate to the entrance of the community." The device quickly responds and plans the optimal route, providing real-time voice guidance while dynamically identifying obstacles, traffic lights, and pedestrian crossings along the route, providing real-time alerts to avoid risks. In indoor scenarios, the device can accurately identify everyday small items such as keys, mobile phones, remote controls, and wallets, providing voice prompts about the location and distance of the items to help users quickly find lost or scattered small items. The device also supports a preset route storage function, allowing users to pre-set frequently used routes such as commuting, shopping, and medical visits. Navigation can be initiated with a single click when traveling, and the system automatically avoids dangerous sections and obstacles.

[0033] At the voice prompt control implementation level, the device has a built-in intelligent priority sorting program that categorizes recognized information into three levels: safety warnings, scene information, and auxiliary services. Safety warnings include high-risk information such as motor vehicles, non-motor vehicles, pedestrians, red lights, steps, and deep pits, which have the highest priority and are broadcast immediately. Scene information includes environmental information such as road signs, door signs, shop names, and traffic signs, which have a secondary priority. Auxiliary services include information such as the location of objects, surrounding facilities, and navigation prompts, which have a normal priority. The broadcast adopts a progressive logic, first warning of the risk in high-risk scenes, and then supplementing the distance, direction, and response methods, such as "Danger ahead, a vehicle is approaching rapidly in 6 meters, please stop and give way." At the same time, users can customize the volume and broadcast interval through buttons or voice to adapt to the usage needs of different user groups such as the elderly and those with mild hearing impairments.

[0034] In terms of privacy and battery life, this device abandons the traditional cloud computing architecture. All image recognition, data processing, and navigation calculations are completed through local edge computing units. It does not collect or upload user travel images, location trajectories, environmental images, or other private data, providing dual protection for user privacy and security from both hardware and algorithm perspectives. The device is equipped with a low-power main control chip, coupled with intelligent sleep and dynamic power consumption adjustment algorithms. It automatically reduces power consumption in standby mode and allocates computing power as needed during operation. A single full charge allows for continuous recognition, broadcasting, and navigation for over 8 hours, fully meeting users' needs for daytime travel and all-day home use.

[0035] In terms of audio output implementation, open-type bone conduction speakers are embedded on both sides of the temples. The sound output position fits the human ear bone and transmits sound through bone conduction. There is no need to block the ear canal. While receiving voice prompts from the device, users can clearly hear ambient sounds such as people's voices, vehicle horns, and intersection prompts. This provides double protection for travel safety and completely solves the safety loophole of traditional guide audio devices that isolate ambient sounds.

Claims

1. A lightweight wearable assistive guide glasses for the blind, characterized in that, include: The main body of the glasses is a lightweight structure with all main hardware integrated inside the frame and temples, and the weight of the whole device is less than or equal to 50 grams. The machine vision recognition module includes a low-power high-definition camera and an AI vision processing unit, equipped with the YOLOv8 algorithm and a lightweight large model, which can recognize static obstacles, dynamic targets and key scene information within a range of 0.5 meters to 10 meters. The intelligent computing and processing module has a built-in local edge computing unit and optical character recognition algorithm, which supports the recognition of various lighting conditions and various fonts, with a recognition accuracy rate of greater than or equal to 97%. The audio prompt module uses an open bone conduction speaker and has a built-in hierarchical priority voice prompt system. The positioning and navigation module integrates a Global Positioning System (GPS) positioning unit and an inertial navigation unit to achieve continuous indoor and outdoor navigation and the search for small items; and A low-power power supply module, which supports continuous operation for more than 8 hours on a single charge.

2. The lightweight wearable assistive guide glasses according to claim 1, characterized in that, The machine vision recognition module identifies static obstacles including steps, roadblocks, walls, and stationary vehicles, and dynamic targets including pedestrians, bicycles, and moving cars. It can also predict the trajectory of dynamic targets and identify collision risks. Key scene information includes traffic lights, pedestrian crossing signs, doorplate text, and facial information.

3. The lightweight wearable assistive guide glasses according to claim 1, characterized in that, The optical character recognition algorithm can adapt to complex lighting environments such as strong light, weak light, and backlight, and is compatible with the recognition of printed and handwritten characters. After recognition, it automatically converts text to speech and broadcasts it in real time.

4. The lightweight wearable assistive guide glasses according to claim 1, characterized in that, The hierarchical priority voice prompt system prioritizes broadcasting critical safety information such as vehicles, red lights, and steps, adopting a progressive broadcast mode of first issuing warnings and then providing detailed explanations, and supports customizable prompt volume and broadcast frequency.

5. The lightweight wearable assistive guide glasses according to claim 1, characterized in that, The positioning and navigation module supports voice interaction and question answering, can search for the location of nearby living facilities, has the function of accurately finding small items such as indoor keys and mobile phones, and also supports preset route obstacle avoidance navigation.

6. The lightweight wearable assistive guide glasses according to claim 1, characterized in that, All identification and computation tasks of the intelligent computing module are completed locally in the edge computing unit, without cloud data upload, thus achieving local protection of user privacy data.

7. A lightweight wearable assistive guide glasses for the blind according to claim 1, characterized in that, The open-type bone conduction speaker is embedded inside the temple of the glasses, fitting snugly against the auricle, without blocking the ear canal or affecting the user's reception of natural ambient sounds.

8. A lightweight wearable assistive guide glasses for the blind according to claim 1, characterized in that, The low-power power supply module, combined with intelligent energy-saving algorithms and a low-power dedicated chip, dynamically adjusts the device's operating power consumption, ensuring continuous operation of the device for extended periods.

Images