This paper presents a scalable memristive neural circuit designed to emulate the spiking and bursting features of crickets' neural circuits for robotic obstacle avoidance. The circuit consists of three memristive Hodgkin-Huxley (HH) neurons, including an ascending neuron that exhibits mixed tonic spiking and bursting features, and two descending neurons that decode these features. The ascending neuron encodes sensing input as mixed firing patterns, while the descending neurons decode these patterns to control the robot's steering and actuation. The circuit is integrated into a robot, demonstrating lower latency compared to conventional platforms, achieving a reduction of more than one order of magnitude. This work paves the way for implementing brain-like systems driven by firing features using memristive neurons, advancing the field of neuromorphic computing and embodied intelligence.This paper presents a scalable memristive neural circuit designed to emulate the spiking and bursting features of crickets' neural circuits for robotic obstacle avoidance. The circuit consists of three memristive Hodgkin-Huxley (HH) neurons, including an ascending neuron that exhibits mixed tonic spiking and bursting features, and two descending neurons that decode these features. The ascending neuron encodes sensing input as mixed firing patterns, while the descending neurons decode these patterns to control the robot's steering and actuation. The circuit is integrated into a robot, demonstrating lower latency compared to conventional platforms, achieving a reduction of more than one order of magnitude. This work paves the way for implementing brain-like systems driven by firing features using memristive neurons, advancing the field of neuromorphic computing and embodied intelligence.