Edge computing models of processing data generated by smart- phones and IoT devices at intermediate edge nodes have shown to reduce communication latency and enhance responsiveness for real-time applications. The past decade has also seen significant developments in processing power, memory, and storage aspects of smartphones. This growth has led to frequent hardware and software upgrades to smartphones, prompting users to discard their old phones for new ones in the market. However, the discarded phones are in perfectly working condition, thereby leaving a rich source of computing power underutilized/untapped. This situation presents an opportunity to harness latent computational resources. This paper introduces MobiTest, a distributed computing framework designed to reuse smartphones as collaborative edge computing nodes. The software framework is developed as an easy plug-and-play type, connecting multiple heterogeneous smartphones in a master slave fashion. A particle swarm optimization-based scheduling algorithm is contextualized and formulated based on the smartphone specifications, including processing capacity, battery status, and device utilization. To validate the framework, we implement and evaluate an object detection task across a distributed set of old smartphones. Results show that our modified optimization algorithm resulted in a 36% and 14% improvement in the CPU utilization when compared with two greedy heuristics. The throughput in terms of data transfer for the task processing also resulted in a 45% and 25% improvement.