BigSnarf blog

Infosec FTW

Speed control for safety

 

 

Note: bouncy image needs vectors stabilized by estimating image-based ego-motion estimation. Then you can measure the speed of objects in the driving window. Horizontal based ego-motions are probably lane dumpers. This estimation could predict lane dumpers and should affect speed and enhanced by other sensors.

http://www.schneidertools.com/wp-content/uploads/2017/01/icpr16-johnny-9.pdf

http://www.frc.ri.cmu.edu/~jizhang03/Publications/ICRA_2015.pdf

  1. What is the distance to each of the objects
  2. What is the speed of the current trajectory?
  3. Optical flow for moving objects
  4. Remove bouncy image with prediction
  5. Filter stationary objects
  6. What objects are left to predict from?
  7. Calculate speed and directions of moving objects
  8. Which objects are travelling with me?
  9. What objects are travelling into my path?

https://github.com/DLuensch/StereoVision-ADCensus

http://www.6d-vision.com/aktuelle-forschung/ego-motion-estimation

http://www.6d-vision.com/9-literatur/keller_dagm11

https://arxiv.org/pdf/1409.7963.pdf

http://cs231n.stanford.edu/reports2016/112_Report.pdf

http://3dvis.ri.cmu.edu/data-sets/localization/

https://github.com/erget/stereovision

https://github.com/JuanTarrio/rebvo

http://www.edwardrosten.com/work/videos/index.html

https://github.com/bigsnarfdude/pykitti

http://www.cvlibs.net/projects.php

http://3dvis.ri.cmu.edu/data-sets/localization/

http://www.robots.ox.ac.uk/NewCollegeData/

http://robots.engin.umich.edu/SoftwareData/Ford

https://www.coursera.org/learn/robotics-perception/lecture/ReEv0/visual-odometry

http://www.cs.toronto.edu/~urtasun/courses/CSC2541/03_odometry.pdf

http://www.6d-vision.com/aktuelle-forschung/ego-motion-estimation

http://www.cvlibs.net/datasets/kitti/eval_odometry.php

https://avisingh599.github.io/vision/visual-odometry-full/

https://en.wikipedia.org/wiki/Visual_odometry

https://sourceforge.net/projects/qcv/

http://docs.opencv.org/3.2.0/d7/d8b/tutorial_py_lucas_kanade.html

http://wiki.ros.org/viso2_ros

https://github.com/uzh-rpg/rpg_svo

 

http://www.eng.auburn.edu/~troppel/courses/00sum13/7970%202013A%20ADvMobRob%20sp13/literature/vis%20odom%20tutor%20part1%20.pdf

 

Optical Flow

 

 

A subject is to provide a pedestrian motion predicting device capable of accurately predicting a possibility of a rush out before a pedestrian actually begins to rush out. According to the embodiments, the pedestrian is detected from input image data, a portion in which the detected pedestrian is imaged is cut out from the image data, a shape of the pedestrian imaged in the cut-out partial image data is classified by collating the shape with a learning-finished identifier group or a pedestrian recognition template group, and the rush out of the pedestrian is predicted based on a result of the acquired classification.

https://ps.is.tuebingen.mpg.de/research_projects/semantic-optical-flow

https://fling.seas.upenn.edu/~xiaowz/dynamic/wordpress/monocap/

 

http://cs.brown.edu/~ls/Publications/SigalEncyclopediaCVdraft.pdf

 

http://www.hizook.com/blog/2010/02/16/learning-estimate-robot-motion-and-find-unexpected-objects-optical-flow

http://www.cc.gatech.edu/~dellaert/pub/Roberts09cvpr.pdf

http://www.vision.cs.ucla.edu/papers/karasevAHS16.pdf

https://arxiv.org/abs/1504.06852

https://arxiv.org/pdf/1503.04036.pdf

https://arxiv.org/abs/1702.05729

Emergency Vehicle Alerting

google-car-1

Emergency Vehicle Alerting system that warns drivers when they are approaching an ambulance, fire engine, police or rescue squad using emergency lights.
A warning system for alerting the driver of a private vehicle that an emergency vehicle is approaching is disclosed. The system includes a receiver and a display panel mounted in the private vehicle, and at least two infrared receivers mounted on the private vehicle. The display panel mounted in the private vehicle including indicating devices that allow the driver of the private vehicle to know of the approaching emergency vehicle as well as the direction to move in order to yield the right of way to an approaching emergency vehicle.

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.399.7544&rep=rep1&type=pdf

http://patents.justia.com/patent/9278689

https://www.google.com/patents/US7057528

https://www.google.com/patents/US20110187559

http://www.google.ca/patents/US7061402

http://www.ifv.nl/kennisplein/Documents/20082-detection-of-emergency-vehicles.pdf

 

Car accident detection – sounds of crashing

hydroplaning noise https://arxiv.org/abs/1511.07035

http://karol.piczak.com/papers/Piczak2015-ESC-ConvNet.pdf

Detecting car part failures with sounds and deep learning http://www.vocativ.com/387520/artificial-intelligence-car/

http://www.cs.tut.fi/sgn/arg/music/tuomasv/parascandolo-icassp2016.pdf

PID Control – SDC Neural Network Cruise Control

CNN – Image RotationInvariance

screen-shot-2017-01-26-at-11-00-41-pm

 

Harmonic Networks: Deep Translation and Rotation Equivariance

 https://arxiv.org/abs/1612.04642v1

Deep Learning is awesome and stupid

road-runner-tunnel-car

We need to look beyond the hype cycle on Deep Learning. Deep Learning is ripe for new discoveries by security researchers.

Deep Learning very popular in recent years. Everyone is talking about Deep Learning and reference AI playing games and beating world champions in Go.

When you call out to Siri or Google for answers, Deep Learning is the technology solving those hard problems on the backend. Deep Learning has taken over image recognition and speech recognition. Self driving cars depend on Deep Learning.

Where do you start? What security controls would you put in place? How do you even secure Deep Learning?

Code and papers below on poisoning the system and creating examples to evade some basic vision systems.

These systems are weak. Securing Deep Learning systems are a green field.

 

17-02-enigma

 

Optical Illusions

http://www.michaelbach.de/ot/

Can’t wait for hackers. We need to look beyond the hype cycle on Deep Learning. Deep Learning is ripe for new discoveries by security researchers.

https://github.com/openai/cleverhans

Aston Martin will focus on cybersecurity before developing a self-driving Lagonda

http://karpathy.github.io/2015/03/30/breaking-convnets/

http://cs.stanford.edu/people/karpathy/break_linear_classifier.ipynb

https://people.eecs.berkeley.edu/~tygar/papers/SML/EECS-2008-43.pdf

https://da-data.blogspot.ca/2017/01/finding-bugs-in-tensorflow-with.html

https://people.eecs.berkeley.edu/~tygar/papers/SML2/Adversarial_AISEC.pdf

https://arxiv.org/pdf/1606.06565.pdf

https://arxiv.org/pdf/1412.6572v3.pdf

https://arxiv.org/pdf/1701.04079v1.pdf

https://arxiv.org/abs/1610.05820

http://vision.stanford.edu/cs598_spring07/papers/Lecun98.pdf

https://arxiv.org/abs/1312.6199

https://arxiv.org/pdf/1412.1897v4.pdf

https://arxiv.org/abs/1609.02943

https://arxiv.org/abs/1610.05820

https://arxiv.org/pdf/1611.01236.pdf

https://people.eecs.berkeley.edu/~tygar/papers/SML/EECS-2008-43.pdf

https://arxiv.org/abs/1611.03814

https://arxiv.org/abs/1611.03814

https://www.endgame.com/blog/endgame-research-aisec-deep-dga

https://conf.startup.ml/blog/adversarial

http://www.slideshare.net/pragroup/secure-kernel-machines-against-evasion-attacks

https://da-data.blogspot.ca/2017/01/finding-bugs-in-tensorflow-with.html

https://arxiv.org/pdf/1606.06565.pdf

https://arxiv.org/pdf/1312.6199v1.pdf

https://arxiv.org/pdf/1701.04079v1.pdf

http://vision.stanford.edu/cs598_spring07/papers/Lecun98.pdf

https://arxiv.org/abs/1312.6199

https://arxiv.org/pdf/1412.1897v4.pdf

http://cacm.acm.org/magazines/2016/11/209133-learning-securely/fulltext

https://arxiv.org/abs/1609.02943

https://arxiv.org/pdf/1412.6572.pdf

https://arxiv.org/abs/1611.03814

https://arxiv.org/abs/1611.03814

https://people.eecs.berkeley.edu/~tygar/papers/SML2/Adversarial_AISEC.pdf

http://composition.al/blog/2016/09/29/thoughts-on-adversarial-examples-in-the-physical-world/

https://conf.startup.ml/blog/adversarial

Main

https://people.eecs.berkeley.edu/~tygar/papers/SML2/Adversarial_AISEC.pdf

https://mascherari.press/introduction-to-adversarial-machine-learning/

https://arxiv.org/pdf/1606.04435v1.pdf

https://www.wired.com/2017/02/hacked-android-phones-unlock-millions-cars/

https://www.ecnmag.com/news/2017/02/cybersecurity-risk-self-driving-cars#.WKclP-zn0xU.twitter

 

 

Agent based self driving car simulator

Backprop

DeepSounds

Style transfer – NIR –> RGB