Say I build gait sequences to move forward, left, right and backwards for a Hexapod. Is there a way to store the sequences in the SSC32 and then call them from the ABB? Like it starts off moving forward, them a sensor detects an object. Can the ABB tell the SSC32 to execute the ‘move left’ gait sequence and it do the rest. Or maybe something like the ABB says ‘forward’ repeatedly until an object is detected then start issuing ‘left’ repeatedly until the object is no longer detected?
It’s possible to store sequences in a EEPROM 24LC256/512 (the empty 8 pin DIL socket on the SSC32 board) and run them with the GP sequencer command. Ref Visual Sequencer manual ver 1.13 page 14.
NB! You have to download The GP Sequencer Firmware : 1-0gbeta.abl.
You COULD, but I’m not sure why. If you’ve got the ABB programmed with one of the samples, you’ve already got a gait engine, and can take inputs for heading and range (X speed and Y speed from joysticks), as well as other parameters.
You’re talking about the 'Bot being Autonomous, of course, so your code’s goal is wandering or exploration, and watching for sensor detections would simply change the stream of parameters being sent to the gait engine.
If sonar detects something in the path, alter the heading (Y speed) to avoid.
If a bump sensor or “cat whisker” detects something, stop forward motion (X speed) and reverse for a short distance, make a slight turn (Y speed), and continue on. Unless, of course, you are in the “follow the wall” behavior! There you’d make slight turns to set your heading to follow the wall.
The reason is less code space is used by the micro. If the micro is maxing out, offloading these sequences is a good way to reclaim memory on a stamp type controller.
Point taken. There are several “canned” sequences that could be put into EEPROM, and the program space reclaimed. But if one already has a gait engine that is parametrically driven, then why not use it? Change the command flow from “forward” to “reverse”, or “veer left”, etc. A few tokens could do that, no need for a “sequence”.
What I’d like to see; and I’m not sure it’s even possible, is to expand gait generation (gait engine) to accommodate some of the postures, or sequences desired. In the process of reversing direction from a collision detection; the 'Bot could “rear up” in an attack posture before retreating or altering course. This would be a “behavior”.
If the gait engine is capable of several gaits, possibly with data supplied from tables (sequence data?), hopefully this can be done. Think of it as “microcodeing” the sequence data. Special gaits (sequences) would thus be parametrically driven/controlled.
Developing sequences might be fun, but I’m more interested in getting some behaviors implemented, and some sensor routines in place (AFTER I get my first hexapod together)! The locomotion I’ve seen in LM’s 'Bots is fine for now. I want to see some INSECT like behavior in my 'Bot!
What is your 'bot platform?
By the way, have you considered using packet for a 'Bot control? Maybe telemetry?
