Sensory impairment, such as vision loss, affects balance, spatial awareness, and independent mobility. Traditional aids like white canes and guide dogs do not fully address hazards at mid-torso or overhead levels, nor do they solve issues related to availability or route instructions. Electronic Travel Aids (ETAs) have expanded capabilities but often compromise on sensing range, high computing and power costs, or ergonomic usability for non-experts. This paper describes an affordable wearable assistive device called VisionBot, which combines ultrasonic obstacle sensors, an ESP32 microcontroller, a DFPlayer Mini voice speaker, and GPS for outdoor use. The firmware is event-driven and non-blocking in nature, and initially verifies the presence of obstacles. It issues audible warnings where danger is imminent and displays turn-by-turn directions (using GPS) where the road is clear. In order to ensure stability of the system, we introduce moving-average filtering and a watchdog. The paper outlines the leading algorithms in audio and GPS prompts, which were all tested on smart glasses, and proves that they are effective in most environments. The future work will improve the robustness of indoor positioning, include a haptic response, and create more energy-efficient perception. VisionBot demonstrates the way in which multimodal tools may be transformed into cheap mobility tools to assist visually impaired people.