Comment by Animats | Hacker News Reader

Top New Best Ask Show Jobs

Comment by Animats | Hacker News Reader

How is this different from all the other companies doing this?

- TeraRanger [1]

- DJI (Phantom) [2]

- Neurala [3]

- Intel / Yuneec [4]

- Aevo [5]

Then there's the work at MIT [6], which is open source (Github code: [7])

[1] http://www.teraranger.com/drone-flies-in-the-forest-with-col... [2] http://www.pcmag.com/article2/0,2817,2500144,00.asp [3] http://www.neurala.com/technology/collision-avoidance-drones... [4] http://www.pcworld.com/article/3019468/ces/intel-demos-colli... [5] http://www.suasnews.com/2015/05/new-collision-avoidance-unit... [6] http://www.roboticstrends.com/article/watch_mit_drone_autono... [7]

The Iris Automation system can see much further and track dynamic objects like aircraft, balloons, birds, etc. This article does a pretty good job of summing up why we need much more than the "bumper solutions" you mention above: https://medium.com/@alexharm/why-situational-awareness-is-ab...

How do they handle tracking dynamic objects if they are also moving dynamically? Is it just a SLAM method, where you can identify objects between frames?

The only way I have seen this done us using post-processing on object edges. There is no horizon, so it would be pretty difficult to do in real time (I.e. ~20 fps), perhaps to slow to make decisions.

The system is much more than just a basic SLAM algorithm, allowing it to have a full picture of the world, tracking both static and dynamic obstacles that the drone may encounter.

The technologies in the OP are also attempting to provide a solution for air traffic collision avoidance. What is the specific use case Iris will solve better than any other existing technology?

Reply on news.ycombinator.com