Abstract: To solve the problem of slow data processing speed, inflexible design and application restriction of intelligent algorithm in domestic single-chip servo control system, a new method was proposed based on an improved multi-neuron neural network algorithm for DSP servo control system of the unmanned reconnaissance vehicle. The multi-neuron PID controller was constructed in a parallel connection and intercross form, and operated in DSP. Experiment and simulation results show that the method can converge quickly and can realize self-adjusting of the system controlled quantity without human intervention, validating the validity of the algorithm. The basic research provides a theoretical basis and a reference for further application of the unmanned reconnaissance vehicle in battlefield reconnaissance and emergency rescue of unexpected disasters.