Terminator Hunter killer Battle Tank

The video above shows T-1-FNY reacting to cars driving past and being told they are not a threat.

 I never used to be into writing code and building my own control systems, I have always been a more traditional builder, however, the T-1 project has seriously opened my eyes not just to how much fun developing control system is but how easy it is to learn. 

Most of my builds are RC using a convensional RC Transmitter and Reciever. 

In building T-1-FNY and moving into this second phase of the build it has been fun giving the robot a sense of autonomy. It has also been a bit scary, as it is very easy to just flow with the technology and not stand back and think about failsafes and consiquenses. What I mean by this is that making T-1-FNY drive forward on her own, after recognizing a colour and moving towards it, is simply a few lines of code and a go button. Problem with this is that 50 percent of the time it will work great but sometimes it does not and it is in these moments that you stop and think how difficult this technology really is in making it fulproof. What I have descovered is that computing power is everything and a small Raspberry Pi, even though it is great, is simply not powerful enough to get to 100 percent. Even the bigger Nvidia Jetson or even larger computers still are not fulproof. To be able to confidently give T-1-FNY autonomous control over her power systems using object recognition as the control triggers I she must be at atleast 99 percent sure that the object she is seeing is that object. I can not allow a 300kg robot autonomous control on the basis that the object she recognized at 46 percent is the target object. 

Anyway what I mean by all this is that using these control systems, object recognition and external gyro control systems is so much fun but it has given me a new found respect for autonomous robot system and how much computing power and great coding they must have to be so smart, also how far we re from a 'real' T-1. 

I have now come to the conclusion that the follow system I was developing for T-1-FNY is now shelved and I an going to concentrate on using the object recognition and external gyro control system to control other parts of the robot, head, arms, rota,lifts, etc and keep drive control fixed to RC. There are still tons of failsafes I need to build in, even in moving her upper body and shoulders as this assembly still weighs in at 80kg. 

Hopefully by the time we launch the T-1 build at the NEC in May I will have lots of these systems up and running and be able to demo them, or atleast T-1-FNY will.