We are trying to accurately detect m/s from a moving vehicle. mmWave gives very mixed results - should we be using Microwave 5.8GHz instead?
Most sensors Libraries are to detect motion in the field (Coming towards or away) not mounted on a moving vehicle. Reflections are giving us mixed results.
Thank you PJ! Where is the Bio? I was pretty sure Musk was not a fan of LIDAR, but I’ll take a look. Really valuable info on the microwave, thank you.
LOL, I know right, He took it out of the current GEN (4) not a fan because it was to $$$,
Originally it was the SHazizzle. But now Thanks to AI i’m sure , the Radar and Video cameras and some serious Software is the current solution.
Here is the Full story…
" Yes, you’re right! Tesla stopped using LIDAR (Light Detection and Ranging) in its cars and instead relies on cameras, radar, and machine learning to power its Autopilot and Full Self-Driving (FSD) features. There are a few key reasons behind this decision:
Cost and Complexity: LIDAR systems are expensive and add complexity to the design of the vehicle. By relying on cameras and radar, Tesla can keep costs lower and avoid adding an additional sensor that might not provide as much long-term value in the context of their vision-based approach.
Vision-Based Approach: Tesla believes that self-driving cars should operate much like humans do, using vision as the primary sensor. Cameras, combined with powerful AI and machine learning algorithms, allow Tesla vehicles to “see” and interpret the world in a way that is similar to how humans do. This means using visual data (from cameras) to identify lanes, obstacles, traffic signs, pedestrians, etc., and processing this information through deep learning.
Redundancy with Radar: While Tesla used to rely on radar for certain situations (like detecting obstacles in low visibility), they have now shifted towards a vision-based system with cameras and neural networks that are trained to handle a variety of driving scenarios. Tesla’s AI software interprets data from these cameras in real time, allowing the car to make decisions and navigate roads more accurately. Elon Musk, Tesla’s CEO, has often mentioned that the company believes a pure vision-based system is sufficient, as it is closer to human-level driving capabilities.
LIDAR’s Limitations: Musk has also criticized LIDAR, calling it a “crutch” for self-driving technology. He believes that LIDAR is not necessary for achieving full autonomy, and that cameras provide much more relevant information when combined with AI. LIDAR can provide 3D spatial data, but it doesn’t have the same level of detail about road signs, traffic lights, or lane markings as cameras can.
Full Self-Driving (FSD) Vision System: Tesla has been increasingly improving its AI-driven vision system, which uses cameras (eight in total, placed around the car) to interpret the surroundings. This system, combined with the powerful onboard computers, allows for features like lane-keeping, automatic lane changes, navigation on highways, and more. Tesla believes that, over time, their neural network and vision system will continue to improve and become more reliable, even without LIDAR.
So, in short, Tesla stopped using LIDAR because they believe that cameras and radar, combined with their advanced AI, are sufficient for achieving the level of autonomy they are aiming for. They also consider this approach more cost-effective, scalable, and aligned with their vision of how self-driving should work.
HTH
GL PJ
So the Watcher with a Wave sensor would be similar
It would work for IN a vehicle, Low , mid and High end… like this but the OP is moving too, either away or Towards the Analyzed Vehicle if I’m reading this right. AI say’s;
Accurately detecting the speed of a moving vehicle (in m/s) depends on the application requirements, such as precision, cost constraints, and operating conditions. Here are some possible solutions, ranging from best accuracy to lowest cost:
High-Accuracy Solutions
GPS Modules:
How It Works: Measures position over time and calculates speed.
Accuracy: ±0.1–0.3 m/s with high-quality GPS modules.
Pros: Easy to integrate, works well in open environments.
Cons: Accuracy decreases in urban canyons, tunnels, or forests.
Wheel Encoders (Vehicle Dependent):
How It Works: Counts wheel rotations and converts them to speed based on wheel circumference.
Accuracy: Very high if calibrated well.
Cost: Varies, starting from $10 for basic optical encoders.
Pros: Independent of environmental factors, ideal for consistent measurements.
Cons: Requires mechanical integration with the vehicle.
Inertial Navigation Systems (INS):
How It Works: Combines accelerometers, gyroscopes, and magnetometers to estimate speed.
Accuracy: High when combined with GPS (sensor fusion).
Cost: $30–$200+.
Pros: Works indoors and outdoors, reliable with sensor fusion.
Cons: Drift issues without GPS correction.
Low-Cost Solutions
MPU6050 or Similar IMU:
How It Works: Integrates acceleration data over time to estimate velocity.
Accuracy: Moderate, but drift accumulates quickly without calibration.
Cost: $2–$5.
Pros: Extremely low cost, easy to integrate.
Cons: Requires advanced filtering (e.g., Kalman filter) and calibration; drift makes it unsuitable for long-term speed tracking.
Hall Effect Sensors:
How It Works: Detects magnetic fields to measure wheel rotation (like an encoder).
Accuracy: High if properly calibrated.
Cost: $2–$10 per sensor.
Pros: Low cost, robust, and durable.
Cons: Mechanical integration required, dependent on the wheel setup.
Smartphone Apps (For Testing):
How It Works: Uses built-in GPS or IMU data from a smartphone.
Accuracy: Varies based on phone quality (±0.5–1.0 m/s).
Cost: Free (requires a smartphone).
Pros: No additional hardware needed for quick testing.
Cons: Limited to the phone’s sensors.
Combination for Best Cost-Effectiveness
For a balance of cost and accuracy:
Use a Low-Cost GPS + IMU Sensor Fusion: A $10 GPS module combined with an MPU6050 IMU can provide accurate speed detection, with the IMU filling in gaps during temporary GPS signal loss.
Wheel Encoder for Precise Applications: If vehicle access is possible, wheel encoders are both low-cost and highly accurate for speed detection.
Recommendation
If cost is the priority, start with a GPS module. If environmental factors (e.g., signal loss) are an issue, augment it with an IMU for sensor fusion. If the setup allows mechanical integration, wheel encoders or Hall effect sensors are excellent low-cost choices.
If the device measuring speed is moving relative to the vehicle being analyzed, the challenge shifts to determining the relative speed between the two objects. Here are some suitable approaches for this scenario, depending on your budget and desired accuracy:
High-Accuracy Options
Doppler Radar Modules:
How It Works: Measures relative speed using the Doppler effect on radio waves reflected off the vehicle.
Accuracy: ±0.2–0.5 m/s depending on the module.
Cost: $20–$100 (e.g., HB100 microwave sensor).
Pros: Directly measures relative speed, works in most weather conditions, no mechanical integration.
Cons: May require signal processing (FFT) and microcontroller integration.
Lidar Sensors: My Favorite
How It Works: Uses laser pulses to measure the time-of-flight or Doppler effect for speed calculation.
Accuracy: ±0.05 m/s or better with high-quality sensors.
Cost: $100–$300 (e.g., Garmin Lidar-Lite).
Pros: Very precise, works in a wide range of environments.
Cons: Higher cost, some susceptibility to environmental factors like heavy rain.
Ultrasonic Sensors with Doppler Effect:
How It Works: Measures changes in frequency of reflected ultrasonic waves to estimate relative speed.
Accuracy: ±0.5 m/s or better in short-range applications.
Cost: $10–$50.
Pros: Low power, cost-effective for close-range.
Cons: Limited range (a few meters), performance varies with ambient noise.
Low-Cost Solutions
Optical Flow Sensors:
How It Works: Tracks movement across a surface (or vehicle) using visual patterns.
Accuracy: Moderate; depends on distance and lighting.
Cost: $20–$50 (e.g., PMW3901 sensor).
Pros: Low-cost and compact.
Cons: Limited range; lighting conditions can affect accuracy.
Camera with Computer Vision:
How It Works: Captures video, and software (e.g., OpenCV) analyzes frames to compute relative speed.
Accuracy: Moderate to high, depending on camera resolution and algorithm.
Cost: $20–$50 for basic cameras, higher for advanced setups.
Pros: Versatile; can detect speed, direction, and distance.
How It Works: Device uses its own GPS and IMU to calculate absolute speed. Subtract the vehicle’s speed (if available via communication or estimate) to determine relative speed.
Accuracy: ±0.3–1.0 m/s.
Cost: $15–$30 (e.g., GPS module + MPU6050 IMU).
Pros: Simple, no direct interaction with the vehicle.
Cons: Dependent on GPS signal quality, additional computational overhead for relative speed calculation.
Best Low-Cost Option
A Doppler radar module (e.g., HB100) is an excellent low-cost and effective choice for directly measuring relative speed.
If radar isn’t feasible, combining GPS and IMU is a versatile alternative, though it requires the ability to infer the vehicle’s speed or assume it is stationary.
Considerations
Environmental Conditions: Doppler radar and lidar work well in all conditions, while optical solutions like cameras or optical flow sensors may struggle in low light or adverse weather.
Range: Ensure the chosen method can cover the necessary distance between the device and vehicle.
Signal Processing: Some methods (e.g., Doppler radar or ultrasonic) require post-processing to extract accurate speed data.
If you’re looking for the best balance of cost and effectiveness, Doppler radar is likely your top choice. It is straightforward to implement and provides accurate, direct relative speed measurements.
PJ 
My Favorite
My brother