The loss of or impairment in vision makes it challenging for blind and visually impaired people (BVIP) to navigate easily in their surroundings. Several solutions were proposed to address this challenge and assist BVIP in navigation by exploiting existing technologies. However, their reliance on pre-installed infrastructure and costly dedicated hardware made them less practical. As an alternative, pedestrian dead reckoning techniques were proposed. However, the slow walking pace of BVIP, the required contact with un-intended obstacles, and the false recognition of activities increase error accumulation, making these techniques less applicable. Therefore, solutions are needed to accurately recognize the walking patterns of BVIP so that efficient navigation solutions can be developed. This article fills this research gap by extending traditional white cane with smartphone sensors. Specifically, a smartphone is used with a conventional white cane to collect data through its sensors on a time-based data window. For smooth recording, a revolving tire is attached at the bottom of the white cane. The collected data is processed by employing the computational resources of the smartphone using our designed app, which identifies the user’s walking patterns such as walking, stairs up/down, sit/stand, and collision. As a case study, these activities were classified using Naïve Bayes, Random Forests, J48, Decision Table, and LibSVM. Among these, Random Forests gave a higher accuracy. These results suggest that the proposed solution is more practical in designing navigation applications for BVIP and may yield better accuracy if tested with more advanced classifiers.