Class MovementDetectorImu

Inheritance Relationships

class MovementDetectorImu : public navtk::inertial::MovementDetectorPlugin 

Accepts inertial data and after an initial stationary calibration period attempts to classify the inertial as moving/not moving in near-real time.

Public Functions

MovementDetectorImu(const Size window = 10, const double calib_time = 30.0)

Constructor.

Parameters

window – Number of IMU measurements to collect and average together before updating status. A longer window can potentially reduce classification error by removing more noise, but will result a longer lag in status (movement status is held constant over the collection period), with a value of 1 indicating real-time estimation.
calib_time – Number of seconds of guaranteed stationary data at the beginning of processing. This stationary data is used to form a rough estimate of sensor biases and noise levels that are used as the basis of movement detection.

Throws

std::invalid_argument – if window is < 1 or calib_time is < 0.

virtual MovementStatus process(not_null<std::shared_ptr<aspn_xtensor::AspnBase>> data) override

Add another IMU measurement and possibly update status.

Parameters: data – IMU measurement to process. Data is stored for various intervals to first allow for the estimation of biases and noise parameters, and then for averaging of measurements. After the initial stationary period, each ‘window’th (as per the constructor argument) call to this function will trigger a new status calculation. Inputs should be sequential in time.
Returns: Best estimate of movement status based on the last full data collection. Value will be constant between every ‘window’th measurement.

virtual aspn_xtensor::TypeTimestamp get_time() override

Time of last measurement fully processed.

Not updated while buffering IMU data to fill buffer windows.

Returns: Last time; TypeTimestamp((int64_t)0) if no measurements yet received.