Millennium Breakthrough — Killer Robots/Sentries/Guards Can Now Collaborate and Fulfill Their Missions
There are 100 human enemy soldiers and tanks trying to infiltrate your territory. Some of them are giving covering fire. Some are attacking your tanks. Some are blocking your forward approach. Some are bombing your supplies behind you.
You! btw are NOT Human! you are just a set of 100 machines, sentry guns. You have to collectively decide how to finish all of them. With minimal collateral damage. With the goal of making sure none of them get past you into your civilian areas.
You are the Last Line of Defense!
There are some terrorists in a 50 floor building. They have 100’s of hostages. they have guns and grenades. They are on different floor. They have a helicopter on the terrace to try and make an exit. They are negotiating for a ransom of $100m. But they will kill the hostages anyway. You are a team of 20 humanoid robots that are supposed to eliminate ALL terrorists and save ALL hostages.
Failure is NOT an option!
You have 3 fighter planes controlled by robots. 1 UAV which is telling you the location of the targets. You know about enemies anti aircraft missiles and radars.
What do you do?
Sounds like two different things.
We don’t know where to start
We don’t know how to design such systems at all
The Robots we’ve been building for decades now are physically perfectly capable of doing a wide range of useful tasks.
But the problem is that we can’t embed them with the needed intelligence to do those things.
So basically creating state-of-the-art robotics is a software challenge ‘Not’ a hardware problem.
*** Ignore the Reinforcement Learning part in this video. Thats not relevant. Thats an approximation/heuristic that may or may not work. And will fail randomly and severely. For the purposes of this article we are dealing exclusively with deterministic algorithms which come with a guarantee, accuracy and can be understood by humans.
1m variables & 5m constraints is actually NOT much when Automatski has the capability of solving problems with 1tn variables and constraints.
But the essential idea is that the Scenarios 1–3 mentioned above can be easily modelled using K-CSP Solvers which are a SUPERSET of K-SAT Solvers.
- Such agents will work asynchronously.
- Will be loosely couples.
- Will coordinate and collaborate to take decisions.
- Work and Solve Constraints
- Optimize and meet their Objectives/Goals
- Will do all this is Near Realtime (as of today, will be in a few milliseconds in a few years)
Win the war for us. Defend us. Kill our enemies…
Now we can give these machines REAL Decision Making, Goal Achieving, Collaborating and Coordinating Power.
Thats our Millennium Breakthrough!
This is our website http://automatski.com