Method for Recognizing Small Obstacles in Assistant Devices for the Blind

The article proposes an innovative method for recognizing obstacles for people with visual impairments, based on the use of two laser range-finder sensors located in parallel at an angle of 45 degrees to the plane of motion. This approach solves the key problem of detecting small objects (curbs, pits, steps) that traditional means such as a white cane or ultrasonic devices often fail to detect due to error caused by the user's hand movement. The presented method eliminates the need for expensive AI solutions, based on the analysis of the distance difference between the sensor readings, taking into account the dynamic correction of the angle of inclination of the device when moving through the gyroscope and accelerometer. A key element of the algorithm is the recognition of six types of patterns characteristic to dangerous obstacles such as pits, curbs, steps and stairs. Two laser sensors are used for detection, the measurement difference of which is recorded in a pattern based on threshold values and timers. At the output of the algorithm, a numerical code is formed, which is compared with the database of obstacle codes. The effectiveness of the method was confirmed by mathematical modeling and full-scale tests, which revealed delays in real sensor data, but retained high detection accuracy. The simplicity of the design and logic of the algorithm makes the solution available for mass use, reducing the risks of social isolation of the visually impaired. The results demonstrate the promise of the method for integration into portable assistive devices that combine reliability and low cost.