Data structure encapsulating sensor model required for Bayesian time-to-impact state estimation. More...
#include <Robots/LoBot/tti/LoSensorModel.H>
Classes | |
| class | Params |
Public Member Functions | |
| SensorModel (const std::string &lgmd_phase) | |
| void | update (float dsigma) |
| float | sigma () const |
| std::string | name () const |
| Return the name of LGMD phase this sensor model corresponds to. | |
| std::vector< float > | column_vector (float lgmd) const |
| std::vector< float > | table () const |
| ~SensorModel () | |
| Clean-up. | |
| int | column_size () const |
| int | row_size () const |
| float | row_min () const |
| float | row_max () const |
| float | row_step () const |
Data structure encapsulating sensor model required for Bayesian time-to-impact state estimation.
This class implements a two-dimensional array (i.e., a table) containing P(lgmd|tti) probability values. The columns of this table correspond to different LGMD ranges; its rows to different TTI states. The array is stored in a single-dimensional array in column-major order.
Definition at line 95 of file LoSensorModel.H.
| lobot::SensorModel::SensorModel | ( | const std::string & | lgmd_phase | ) |
As mentioned earlier, the LGMD input signal is partitioned into two phases, viz., LOOMING and BLANKING. Robolocust uses different causal likelihood profiles for these phases. Thus, when creating a sensor model object, the client must specify the phase corresponding to this instance so that it can retrieve the correct set of parameters from the config file.
Definition at line 91 of file LoSensorModel.C.
References column_size(), R, and update().
| lobot::SensorModel::~SensorModel | ( | ) |
Clean-up.
Definition at line 214 of file LoSensorModel.C.
| int lobot::SensorModel::column_size | ( | ) | const [inline] |
As mentioned earlier, the sensor model is a table containing causal probabilities. These methods return the sizes of this table's rows and columns and the its total size.
Definition at line 239 of file LoSensorModel.H.
Referenced by column_vector(), SensorModel(), and lobot::TTIEstimator::TTIEstimator().
| std::vector< float > lobot::SensorModel::column_vector | ( | float | lgmd | ) | const |
This method returns an STL vector containing the column vector "pointed to" by the given LGMD value.
Definition at line 197 of file LoSensorModel.C.
References column_size().
Referenced by lobot::TTIEstimator::update().
| std::string lobot::SensorModel::name | ( | ) | const [inline] |
Return the name of LGMD phase this sensor model corresponds to.
Definition at line 257 of file LoSensorModel.H.
| float lobot::SensorModel::row_min | ( | ) | const [inline] |
These methods return the minimum, maximum and step values associated with the discretization used on the rows (i.e., TTI values).
Definition at line 247 of file LoSensorModel.H.
Referenced by lobot::TTIEstimator::update().
| float lobot::SensorModel::sigma | ( | ) | const [inline] |
Return the current value of the standard deviation used for the Gaussian weighting for neighbouring bins.
Definition at line 254 of file LoSensorModel.H.
| std::vector< float > lobot::SensorModel::table | ( | ) | const |
This method returns an STL vector containing the entire table of probabilities.
Definition at line 206 of file LoSensorModel.C.
| void lobot::SensorModel::update | ( | float | dsigma | ) |
This method can be used to recompute the sensor model using some new value for the standard deviation of the Gaussian weighting formula used for the increment factors of each bin in the sensor model's probability table.
NOTE: The standard deviation provided to this function is actually an increment factor for the current sigma value maintained internally by the sensor model object. Thus, the dsigma parameter to this function will be added to the m_sigma member variable. This design caters to a client behaviour that is specifically designed to update and visualize the available sensor models.
Definition at line 136 of file LoSensorModel.C.
References R.
Referenced by SensorModel().
1.6.3