Olzhas ADIYATOV olzhas

VirtualGL

When you use ssh with X forwarding, you might have noticed that you cannot execute programs that require 3D acceleration. That's where VirtualGL comes into play.

To forward X11 from inside a docker container to a host running macOS

Install XQuartz: https://www.xquartz.org/
Launch XQuartz. Under the XQuartz menu, select Preferences
Go to the security tab and ensure "Allow connections from network clients" is checked.
Run xhost + ${hostname} to allow connections to the macOS host *
Setup a HOSTNAME env var export HOSTNAME=`hostname`*
Add the following to your docker-compose:

 environment:

Turn off all swap processes

sudo swapoff -a

Resize the swap

sudo dd if=/dev/zero of=/swapfile bs=1G count=8

if = input file
of = output file

The document provides description on calibration of three Kinect for Microsoft sensors connected to one computer with several usb controllers. Three cameras setup is shown below:

Intrinsic, extrinsic, and Kinect2Kinect calibration is performed to know the position of each sensor in the space. Our setup is ROS Indigo with Ubuntu 14.04. freenect_launch and camera_pose ROS packages are used. Camera_pose package provides the pipeline to calibrate the relative 6D poses between multiple camera's. freenect_launch package contains launch files for using OpenNI-compliant devices in ROS. It creates a nodelet graph to transform raw data from the device driver into point clouds, disparity images, and other products suitable for processing and visualization. It is installed with catkin as follows:

# Prepa

CPU governer matters

If you need increase in computational power from your workstation this might be a right post for you. Power management, in this case CPU Governor, is a software that assures the "best" power consumption by a computer for a particular task. For some reason in Ubuntu 20.04 a default CPU governor (even on stationary PCs) is powersave. This comes in handy, if you are using it for some office applications (text processor, email, client, webbrowsing). However for a real-time control of the electromechanical systems such as robots where we have to be able to use computers performance at full, or for machine learning we desire to minimize time spent on training a model. A small amount of performance boost can be obtain choosing the right CPU governor in our case performance profile. You can enable performance CPU governor on your computer following the instructions below. Open your favorite shell and type following command.

$ sudo systemctl disable ondemand.service
$ echo perfor

CMake

target_link_library(something private LIB)

The thing above is only applicable for shared libraries

	#ifndef TIMER_H
	#define TIMER_H

	#include <chrono>
	#include <thread>

	using timer = std::chrono::high_resolution_clock::time_point;

	inline void tic(timer &t){
	t = std::chrono::high_resolution_clock::now();

	import numpy as NP
	import casadi as C
	def qpsolve(H,g,lbx,ubx,A=NP.zeros((0,0)),lba=NP.zeros(0),uba=NP.zeros(0)):
	# Convert to CasADi types
	H = C.DMatrix(H)
	g = C.DMatrix(g)
	lbx = C.DMatrix(lbx)
	ubx = C.DMatrix(ubx)
	A = C.DMatrix(A)
	A = A.reshape((A.size1(),H.size1())) # Make sure matching dimensions

	sudo update-alternatives --install /usr/bin/clang++ clang++ /usr/bin/clang++-5.0 100
	sudo update-alternatives --install /usr/bin/clang clang /usr/bin/clang-5.0 100