More Filter Changes by jpanikulam · Pull Request #108 · build-week/hover-jet

jpanikulam · 2019-03-24T18:43:59Z

No description provided.

IJDykeman · 2019-04-14T14:23:35Z

filtering/filter_balsaq.cc

+  //   const auto time_of_validity = to_time_point(imu_msg.timestamp);
+  //   const jcc::Vec3 gyro_radps(imu_msg.gyro_radps_x, imu_msg.gyro_radps_y, imu_msg.gyro_radps_z);
+  //   estimation::jet_filter::GyroMeasurement gyro_meas;
+  //   gyro_meas.observed_w = gyro_radps;


IJDykeman · 2019-04-14T14:24:04Z

filtering/filter_viz_balsaq.cc

+      sensor_geo_->add_ellipsoid({fit.ellipse, jcc::Vec4(0.4, 0.6, 0.4, 0.7), 2.0});
+      sensor_geo_->flush();
+    };
+    // std::cout << "Optimizing" << std::endl;


// std::cout << screams

IJDykeman · 2019-04-27T15:48:43Z

filtering/filter_demo.cc

+
+    // speeds_plot.subplots["opt_est_dot_x"].points.push_back({dt, x.x.R_world_from_body.log().x()});
+    // speeds_plot.subplots["opt_est_dot_y"].points.push_back({dt, x.x.R_world_from_body.log().y()});
+    // speeds_plot.subplots["opt_est_dot_z"].points.push_back({dt, x.x.R_world_from_body.log().z()});


remove comments before landing

IJDykeman

will review the rest later

IJDykeman · 2019-07-09T03:08:00Z

camera/cfg/camera_config_1.yaml

 note: these cal values were copied from camera 0
 camera_matrix: [499.7749869454186, 0, 309.3792706235992,
-                0, 496.9300965132637, 241.6934416030273, 
+                0, 496.9300965132637, 241.6934416030273,


IJDykeman · 2019-07-09T03:10:48Z

control/controller_balsaq.cc


+  const jcc::Vec3 w_jet_frame = pose.world_from_jet.inverse() * pose.w_world_frame;
+
+  const MatNd<3, 3> Kp = cfg.kp.asDiagonal();


3.6 roentgen/hr on this abbreviation

IJDykeman · 2019-07-09T03:12:55Z

control/controller_balsaq.hh

 namespace control {

+struct GainConfig {
+  jcc::Vec3 kp;


expand acronym

IJDykeman · 2019-07-09T03:43:53Z

filtering/calibrate_camera_from_log.cc

+      node["camera_matrix"].push_back(K.at<double>(i, j));
+    }
+  }
+  node["resolution"].push_back(cols);


low nobility factor maintaining this variable as state while iterating over images.

Wait -- what do you mean?

referring to the fact that we set cols = image->image.cols; above as a way of getting the resolution for all the images. Could improve by either giving a more specific name so that once doesn't have to scroll way up to see what it refers to. Or by breaking all the image reading into a separate function that returns a struct that keeps all the data you get from reading images together.

IJDykeman · 2019-07-09T03:45:15Z

filtering/calibrate_camera_from_log.cc

+  jcc::Warning() << "Calibrating camera, this might take a while..." << std::endl;
+  int flag = 0;
+  const auto t0 = jcc::now();
+  cv::calibrateCamera(object_points, image_points, cv::Size(480, 270), K, D, rvecs, tvecs, flag);


Pull out resolution constant

IJDykeman · 2019-07-09T04:07:03Z

filtering/calibrate_jet.cc

+}
+
+jcc::Optional<ejf::FiducialMeasurement> find_nearest_fiducial(
+    const std::vector<estimation::TimedMeasurement<ejf::FiducialMeasurement>> fiducial_meas,


meas =/= measurements

IJDykeman · 2019-07-09T04:07:29Z

filtering/calibrate_jet.cc

+  std::cout << node << std::endl;
+}
+
+jcc::Optional<ejf::FiducialMeasurement> find_nearest_fiducial(


call it eg find nearest in time

IJDykeman · 2019-07-09T04:24:22Z

filtering/calibrate_jet.cc

+  return SE3(world_from_camera, jcc::Vec3::Zero());
+}
+
+ejf::Parameters compute_parameters(


better name than compute_parameters

IJDykeman · 2019-08-20T21:23:17Z

control/jet_vane_model.hh


-  double gear_ratio_servo_from_vane = 5.33;
+  // double gear_ratio_servo_from_vane = 5.33;
+  double gear_ratio_servo_from_vane = 1.0;


This doesn't appear to be the actual ratio. Are we sure we should land this?

IJDykeman · 2019-08-20T21:26:03Z

embedded/servo_bq/servo_balsaq.cc

 void ServoBq::init(const Config& config) {
-  subscriber = make_subscriber("servo_command_channel");
+  subscriber_ = make_subscriber("servo_command_channel");
+  publisher_ = make_publisher("servo_pwm_commands_channel");


I might indicate in the subscriber and publisher names the channels they are on.

IJDykeman · 2019-08-20T21:27:38Z

embedded/servo_driver/servo_driver.cc

-void ServoDriver::set_percentage(float unchecked_percentage) {
-  if (unchecked_percentage > 100 || unchecked_percentage < 0) {
+void ServoDriver::set_percentage(const double unchecked_percentage, const int max_pwm_count, const int min_pwm_count) {
+  if (unchecked_percentage > 1.0 || unchecked_percentage < 0.0) {


It looks like you are calling this "percentage" but it is expected to be in [0,1] not [0,100]. Rename?

IJDykeman · 2019-08-20T21:30:36Z

filtering/calibrate_camera_from_log.cc

+  // Set up the viewer
+  //
+
+  setup();


call it eg setup_viewer

IJDykeman · 2019-08-20T21:37:51Z

filtering/calibrate_camera_from_log.cc

+  constexpr int DECIMATION = 25;
+
+  int i = 0;
+  while (true) {


might break this into a function that returns in the data you need from the images.

IJDykeman · 2019-08-20T21:38:25Z

filtering/calibrate_camera_from_log.cc

+      node["camera_matrix"].push_back(K.at<double>(i, j));
+    }
+  }
+  node["resolution"].push_back(cols);


referring to the fact that we set cols = image->image.cols; above as a way of getting the resolution for all the images. Could improve by either giving a more specific name so that once doesn't have to scroll way up to see what it refers to. Or by breaking all the image reading into a separate function that returns a struct that keeps all the data you get from reading images together.

IJDykeman · 2019-08-21T02:31:18Z

filtering/calibrate_jet.cc

+    jf.measure_fiducial(fiducial_meas.measurement, fiducial_meas.timestamp);
+  }
+
+  constexpr bool USE_ACCELEROMETER = false;


Seems too important to be floating down here.

IJDykeman · 2019-08-21T02:47:42Z

filtering/calibrate_jet.cc

+      collected_fiducial_meas.push_back({fiducial_measurement, t});
+    }
+
+    // TODO: NEXT, FILL THIS IN


Stray comment?

IJDykeman · 2019-08-21T02:51:29Z

filtering/extract_data_from_log.cc

+
+  const std::vector<std::string> imu_channels = one_imu_present ? single_imu : both_imu;
+
+  // const bool two_imus = !one_imu_present;


IJDykeman · 2019-08-21T02:53:05Z

filtering/filter_balsaq.cc

+  gonogo().nogo("Starting up");
+  std::cout << "Starting up" << std::endl;
+
+  std::cout << "Subscribing Fiducial..." << std::endl;


Remove prints

IJDykeman · 2019-08-21T02:57:42Z

filtering/filter_viz_balsaq.cc

+
+    pose_geo_->add_axes({pose.world_from_jet, 0.55, 3.0, true});
+
+    // const jcc::Vec3 jet_origin(1.0, 0.0, 0.0);


Delete comments

IJDykeman · 2019-08-21T02:59:49Z

filtering/filter_viz_balsaq.cc

    pose_geo_->flip();
+
+    const SE3 body_from_model = jcc::exp_z(-M_PI * 0.5);
+    // const SE3 body_from_model = jcc::exp_z(0.0);


Delete comment

IJDykeman · 2019-08-21T03:01:04Z

filtering/pose_message.hh


+//
+// Pose has the following contract:
+//   - Z points in the gravity direction


Might disambiguate by writing +Z here since it looks like -Z by accident.

IJDykeman · 2019-08-21T03:02:05Z

infrastructure/comms/mqtt_subscriber.cc

-    std::cerr << channel_name_
-              << ": Not connected. Skipping read. Reason: " << (connecting_ ? "Connecting" : "Not connected")
-              << std::endl;
+    // std::cerr << channel_name_


Remove this and the one above

jpanikulam requested review from IJDykeman and benkroop March 24, 2019 18:43

jpanikulam force-pushed the more_filter_world branch 3 times, most recently from 1390815 to bf1ab95 Compare March 30, 2019 18:21

jpanikulam mentioned this pull request Apr 5, 2019

Changes to permit new opencv version #82

Open

IJDykeman reviewed Apr 14, 2019

View reviewed changes

IJDykeman reviewed Apr 27, 2019

View reviewed changes

jpanikulam force-pushed the more_filter_world branch 3 times, most recently from 9438487 to 7b2d67c Compare July 7, 2019 23:15

IJDykeman reviewed Jul 9, 2019

View reviewed changes

masonturner and others added 9 commits July 13, 2019 21:54

Rewrite Servo Driver

752a32b

Add pwm to comment

70f3049

Finish calibrations

f843ecf

Rename to reflect target angles

bbd06c2

Address Jake Comments

ea219cb

Add publisher for commanded PWM set points

31387bb

Fix Servo Test Target

e9eddb4

Updates to filtering

9e8965d

Fix first round of merge conflicts

cb0c985

jpanikulam force-pushed the more_filter_world branch from c380b4c to cb0c985 Compare July 14, 2019 01:55

jpanikulam added 4 commits July 14, 2019 01:10

Debug prints

281b134

Update experiments

f9eb8fd

Updates

d2ba65d

Push latest

f0d4063

IJDykeman reviewed Aug 21, 2019

View reviewed changes

IJDykeman closed this Aug 21, 2019

IJDykeman reopened this Aug 21, 2019

IJDykeman self-requested a review August 21, 2019 03:28

IJDykeman approved these changes Aug 21, 2019

View reviewed changes


		const jcc::Vec3 w_jet_frame = pose.world_from_jet.inverse() * pose.w_world_frame;

		const MatNd<3, 3> Kp = cfg.kp.asDiagonal();


		const std::vector<std::string> imu_channels = one_imu_present ? single_imu : both_imu;

		// const bool two_imus = !one_imu_present;


		pose_geo_->add_axes({pose.world_from_jet, 0.55, 3.0, true});

		// const jcc::Vec3 jet_origin(1.0, 0.0, 0.0);

Conversation

jpanikulam commented Mar 24, 2019

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

IJDykeman left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants