vsunspiral
commented
9 years ago well, in the ideal sense, we would want different noise models (distribution, scale, etc) available so that appropriate ones could be used on a per-sensor / per-actuator basis -- i.e. some sensors may have a 1% noise, and some sensors may have 5% noise with a specific bias.
I would think that the right balance would be to have a default noise model that is used if you just select "inject noise", yet one could override specific sources of noise with specific models.
after setting up all the noise models, it would be good if one could turn it on/off by config setting.
vytas
On Jan 12, 2015, at 8:11 AM, Brian Mirletz notifications@github.com wrote:
In order to test the robustness of controllers, we should add options for sensor/actuator noise. I can think of at least three options:
1) Noise in initial lengths/pretension
2) Noise in sensor data (length, tension on actuators, others for other sensors)
3) Noise in the results of commands
Feature 3 was implemented in V 0.1 and not brought forward, but all three should be a fairly easy subclasses of tgBasicActuator and tgKinematicActuator.
Thoughts on other types of noise, and the degree of noise?
— Reply to this email directly or view it on GitHub.
Vytas SunSpiral
Dynamic Tensegrity Robotics Lab
cell- 510-847-4600
Office: 650-604-4363
N269 Rm. 100
Stinger Ghaffarian Technologies Intelligent Robotics Group NASA Ames Research Center
I will not tiptoe cautiously through life only to arrive safely at death.
In order to test the robustness of controllers, we should add options for sensor/actuator noise. I can think of at least three options:
1) Noise in initial lengths/pretension
2) Noise in sensor data (length, tension on actuators, others for other sensors)
3) Noise in the results of commands
Feature 3 was implemented in V 0.1 and not brought forward, but all three should be a fairly easy subclasses of tgBasicActuator and tgKinematicActuator.
Thoughts on other types of noise, and the degree of noise?