Skip to content

Instantly share code, notes, and snippets.

View madratman's full-sized avatar

Ratnesh Madaan madratman

138 followers · 149 following
View GitHub Profile
@madratman
madratman / airsim_neurips_reset_episode_crash_repro.py
Last active November 15, 2019 00:45
https://github.com/microsoft/AirSim-NeurIPS2019-Drone-Racing/issues/123
import airsimneurips
import threading
import time
import random
class ReproduceResetRaceCondition():
def __init__(self, drone_name = "drone_1"):
self.airsim_client = airsimneurips.MultirotorClient()
self.airsim_client_2 = airsimneurips.MultirotorClient()
self.airsim_client_3 = airsimneurips.MultirotorClient()
@madratman
madratman / airsim_neurips_god_settings_nodisplay.json
Created November 5, 2019 21:27
{
"ClockSpeed": 1,
"SeeDocsAt": "https://github.com/Microsoft/AirSim/blob/master/docs/settings.md",
"SettingsVersion": 1.2,
"SimMode": "Multirotor",
"ViewMode": "NoDisplay",
"Vehicles": {
"drone_1": {
"Cameras": {
"fpv_cam": {
@madratman
madratman / unreal_windows_cross_compile_linux.md
Last active August 7, 2019 21:42
@madratman
madratman / gif_from_mp4.md
Created July 15, 2019 22:30 
mkdir frames
ffmpeg -i video.mp4  -r 5 'frames/frame-%05d.png'
mogrify -resize 360x640 *.png
convert -delay 10 -loop 0 *.png myimage.gif
@madratman
madratman / unreal_airsim_bug_with_multiple_displays_xinerama_enabled_xrandr_not_found.txt
Created February 25, 2019 19:34
madratman@madratman-HP-Z4-G4-Workstation:~$ ./Downloads/Blocks/Blocks.sh -windowed
Increasing per-process limit of core file size to infinity.
LogPlatformFile: Using cached read wrapper
LogInit: Display: RandInit(754615442) SRandInit(754615442).
LogTaskGraph: Started task graph with 5 named threads and 17 total threads with 3 sets of task threads.
LogStats: Stats thread started at 0.022213
LogPluginManager: Mounting plugin AirSim
LogPluginManager: Mounting plugin PhysXVehicles
LogPluginManager: Mounting plugin Paper2D
LogPluginManager: Mounting plugin LightPropagationVolume
@madratman
madratman / owl_lie_algebra_meme.tex
Last active March 28, 2018 01:32
% Totally not inspired by LSD-SLAM https://vision.in.tum.de/_media/spezial/bib/engel14eccv.pdf
\documentclass{article}
\usepackage[utf8]{inputenc}
\usepackage{amsmath}
\usepackage{amsfonts}
\begin{document}
A 3D rigid body transform, $\mathbf{G} \in SE(3)$ :
\begin{flalign*}
&\mathbf{G} =
@madratman
madratman / ompl_ros_rigidbodyplanning.cpp
Created November 19, 2017 06:54
/*********************************************************************
* Software License Agreement (BSD License)
*
* Copyright (c) 2010, Rice University
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
@madratman
madratman / cluster_simple.md
Created November 16, 2017 05:45
No BS cluster instructions
  • Login to cluster ssh rohitgarg617@perceptron.ri.cmu.edu

Request a node (use any one of these) srun -w clamps -t 5-00:00 -p gpu --gres=gpu:1 --pty /bin/bash srun -w roberto -t 5-00:00 -p gpu --gres=gpu:1 --pty /bin/bash

  • to run your docker image: nvidia-docker run -it --rm --ipc=host -e CUDA_VISIBLE_DEVICES=`echo $CUDA_VISIBLE_DEVICES` -v /data/datasets:/data/datasets -v /storage2/datasets:/storage2/datasets -v /local:/local -v /home/$USER:/home/$USER -v /storage1:/storage1 nvidia/cuda:8.0-cudnn5-devel-ubuntu16.04
@madratman
madratman / slzeng_do_this_shiz.py
Created October 31, 2017 20:47
import torch
import pytorch_autoencoder
list_of_numbers = get them from cpp
sum = sum
model = pytorch_autoencoder.ConvCVAENet(state_dim = 2,
condition_dim = (200, 200, 2),
latent_dim = 3,
encoder_hidden_dim = 512,
@madratman
madratman / keybase.md
Created October 9, 2017 22:02

Keybase proof

I hereby claim:

  • I am madratman on github.
  • I am madratman (https://keybase.io/madratman) on keybase.
  • I have a public key ASAh2qBRrxv6r_SqPJErkN6vlZkDayjsVy5hGdtSu1sl2wo

To claim this, I am signing this object: