This article introduces M-detector, a novel method for moving event detection using LiDAR point streams. Unlike traditional methods that accumulate LiDAR points into frames and detect object-level motions, M-detector determines if a point is moving immediately after its arrival, achieving a point-by-point detection with a latency of just several microseconds. M-detector is based on the occlusion principle and can be used in various environments with different types of LiDAR sensors. The method is effective on various datasets and applications, showcasing superior accuracy, computational efficiency, detection latency, and generalization ability. Autonomous robots, including self-driving vehicles and drones, have the potential to revolutionize applications such as last-mile delivery, robotaxi, agriculture, and aerial imaging. However, one of the key challenges in deploying these robots is detecting and avoiding non-cooperative moving objects. M-detector addresses this challenge by detecting fast-moving objects or their moving parts immediately after the motion occurs. This task is known as moving event detection or event detection. Event detection is typically achieved using event cameras, which detect changes in a scene with a reaction time of microseconds. Unlike traditional cameras, event cameras measure the change in intensity of a pixel rather than the intensity itself. This generates a stream of asynchronous events with microsecond-level latency. Due to their high dynamic range, low power consumption, and low detection latency, event cameras have been used in various applications such as dynamic obstacle avoidance for quadrotors, video reconstruction in high-speed motion, and visual inertial odometry for extreme motion conditions. LiDAR sensors are another type of sensor widely used for autonomous robots. Unlike cameras, LiDAR sensors measure the depth of a pixel by emitting a laser pulse along the pixel direction and computing the laser time of flight (ToF). This active and direct ranging mechanism can produce depth measurements that are very accurate, efficient, and independent of external illumination. LiDAR sensors often have tens to hundreds of laser emitters stacked in an array and each one emits laser pulses at microseconds interval, producing tens of thousands to millions of points per second. These high-frequency point measurements, although at a fixed rate, have a temporal resolution of microseconds to sub-microseconds that is similar to event cameras. Fully exploiting these high-frequency measurements could provide extremely timely detection of any moving events in the scene. Specifically, it requires a moving point to be detected immediately after its arrival to minimize the latency. This online detection of moving events at the rate of point sampling is analogous to event cameras and hence referred to as event detection. Moving event detection could be achieved at the measuring stage of a LiDAR sensor, such as the Frequency-Modulated Continuous Wave Laser Detection and Ranging (FMCW-LADAR) sensors. Compared to the standard ToF LiDAR, FMCW-LADAR involves a continuously emitted laser beam and utilizes the Doppler effect to acquire information on range and velocity. While being able to measure theThis article introduces M-detector, a novel method for moving event detection using LiDAR point streams. Unlike traditional methods that accumulate LiDAR points into frames and detect object-level motions, M-detector determines if a point is moving immediately after its arrival, achieving a point-by-point detection with a latency of just several microseconds. M-detector is based on the occlusion principle and can be used in various environments with different types of LiDAR sensors. The method is effective on various datasets and applications, showcasing superior accuracy, computational efficiency, detection latency, and generalization ability. Autonomous robots, including self-driving vehicles and drones, have the potential to revolutionize applications such as last-mile delivery, robotaxi, agriculture, and aerial imaging. However, one of the key challenges in deploying these robots is detecting and avoiding non-cooperative moving objects. M-detector addresses this challenge by detecting fast-moving objects or their moving parts immediately after the motion occurs. This task is known as moving event detection or event detection. Event detection is typically achieved using event cameras, which detect changes in a scene with a reaction time of microseconds. Unlike traditional cameras, event cameras measure the change in intensity of a pixel rather than the intensity itself. This generates a stream of asynchronous events with microsecond-level latency. Due to their high dynamic range, low power consumption, and low detection latency, event cameras have been used in various applications such as dynamic obstacle avoidance for quadrotors, video reconstruction in high-speed motion, and visual inertial odometry for extreme motion conditions. LiDAR sensors are another type of sensor widely used for autonomous robots. Unlike cameras, LiDAR sensors measure the depth of a pixel by emitting a laser pulse along the pixel direction and computing the laser time of flight (ToF). This active and direct ranging mechanism can produce depth measurements that are very accurate, efficient, and independent of external illumination. LiDAR sensors often have tens to hundreds of laser emitters stacked in an array and each one emits laser pulses at microseconds interval, producing tens of thousands to millions of points per second. These high-frequency point measurements, although at a fixed rate, have a temporal resolution of microseconds to sub-microseconds that is similar to event cameras. Fully exploiting these high-frequency measurements could provide extremely timely detection of any moving events in the scene. Specifically, it requires a moving point to be detected immediately after its arrival to minimize the latency. This online detection of moving events at the rate of point sampling is analogous to event cameras and hence referred to as event detection. Moving event detection could be achieved at the measuring stage of a LiDAR sensor, such as the Frequency-Modulated Continuous Wave Laser Detection and Ranging (FMCW-LADAR) sensors. Compared to the standard ToF LiDAR, FMCW-LADAR involves a continuously emitted laser beam and utilizes the Doppler effect to acquire information on range and velocity. While being able to measure the