This paper presents the design of a low-cost assistive system for people with visual impairments, aimed at improving obstacle detection through perception technologies and embedded processing. The objective is to develop a system integrating complementary sensors and multisensory outputs to facilitate the safe mobility in real-world environments. The system comprises two subsystems: a primary one based on an NVIDIA Jetson Nano unit that processes data from an RPLIDAR A1 sensor and a computer vision camera; and an auxiliary one using Arduino and JSN-SR04T ultrasonic sensors, which provides tactile alerts via vibration motors. The proposal was structured through problem identification, requirements definition, component selection, and the design of a functional block-based architecture. The system is expected to detect static and dynamic obstacles at distances of up to 1.5 m, with real-time responses from the auxiliary module at ranges below 70 cm. It records variables such as distance, alert type, and response time. The proposed system represents a significant step toward developing accessible and modular solutions for people with visual impairments. It is intended as a foundation for future user testing to assess its impact, usability, and feasibility for implementation in both indoor and outdoor public environments.