Mercurial Hosting > traffic-intelligence
view scripts/safety-analysis.py @ 348:c64a4f889b97
added safety analysis options to feature tracking (with default values for backward compatibility)
author | Nicolas Saunier <nicolas.saunier@polymtl.ca> |
---|---|
date | Wed, 26 Jun 2013 18:40:31 -0400 |
parents | 14a2405f54f8 |
children | 7e9ad2d9d79c |
line wrap: on
line source
#! /usr/bin/env python import utils, storage, prediction, events import sys, argparse import matplotlib.pyplot as plt import numpy as np parser = argparse.ArgumentParser(description='The program processes indicators for all pairs of road users in the scene') parser.add_argument('configFilename', help = 'name of the configuration file') #parser.add_argument('--maxspeed', dest = 'maxSpeed', help = 'maximum speed when predicting future motion (km/h)', default = 50, type = int) #parser.add_argument('--time-horizon', dest = 'maxSpeed', help = 'maximum speed when predicting future motion (km/h)', default = 50, type = int) args = parser.parse_args() # TODO work on the way to indicate an interaction definition # if False: # test if there is a configuration file? params = utils.TrackingParameters() params.loadConfigFile(args.configFilename) # configuration parameters # TODO from command line maxSpeed = args.maxSpeed/3.6/params.videoFrameRate # speed limit 50 km/h for urban envt, 90km/hr = 25 m/sec for highways timeHorizon= params.videoFrameRate*5 # prediction time Horizon = 1.5 s (reaction time) (5 second) collisionDistanceThreshold= 1.8 # m computeCZ = False # display = False # parameters for prediction methods constantVelocityPredictionParameters = prediction.ConstantPredictionParameters(maxSpeed) normalAdaptationPredictionParameters = prediction.NormalAdaptationPredictionParameters(maxSpeed, 100, 2./frameRate**2, # m/s2 0.2/frameRate) # rad/s featurePredictionParameters = prediction.PointSetPredictionParameters(maxSpeed) evasiveActionPredictionParameters = prediction.EvasiveActionPredictionParameters(maxSpeed, 100, -9.1/frameRate**2, # m/s2 4.3/frameRate**2, # m/s2 0.5/frameRate, # rad/s False) featureEvasiveActionPredictionParameters = prediction.EvasiveActionPredictionParameters(maxSpeed, 10, -9.1/frameRate**2, # m/s2 4.3/frameRate**2, # m/s2 0.5/frameRate, # rad/s True) objects = storage.loadTrajectoriesFromSqlite(params.databaseFilename,'object') # features = storage.loadTrajectoriesFromSqlite('amherst-10.sqlite','feature') # needed if normal adaptation interactions = events.createInteractions(objects) for inter in interactions: inter.computeIndicators() inter.computeCrossingsCollisions(constantVelocityPredictionParameters, collisionDistanceThreshold, timeHorizon, computeCZ) storage.saveIndicators(params.databaseFilename, interactions) # if display: # plt.figure() # plt.axis('equal') # for inter in interactions[:2]: # for collisionPoints in inter.collisionPoints.values(): # for cp in collisionPoints: # plot([cp.x], [cp.y], 'x') # for the demo, output automatically a map # possibility to process longitudinal coords only