Autonomous mobile robots operating in indoor environments must be capable of avoiding humans as dynamic obstacles while maintaining reliable navigation performance. Although the Vector Field Histogram (VFH) algorithm has been widely applied for obstacle avoidance, most previous studies rely on two-dimensional sensors or do not directly integrate three-dimensional point cloud perception with real-time navigation on differential drive mobile robots. This limitation motivates the development of a more robust human-aware navigation system. Therefore, this study proposes a Differential Drive Mobile Robot (DDMR) that integrates Microsoft Kinect V1 point cloud sensing with the Vector Field Histogram (VFH) algorithm for real-time collision avoidance. Environmental depth information is converted into a three-dimensional point cloud, processed into a VFH polar histogram, and used to determine safe navigation directions during robot motion. Experimental evaluation was conducted in dynamic indoor environments involving human obstacles. The proposed system processes point cloud data at approximately 22 FPS, detects humans within a range of 0.8–3.5 m with a measurement error of 0.042 m, achieves an average VFH computation time of 0.05 s, and produces an average avoidance response time of 0.73 s with a collision-free success rate of 93.3%. These results demonstrate that integrating Kinect-based point cloud perception with VFH provides an effective and practical solution for real-time human-aware obstacle avoidance in indoor autonomous mobile robot navigation