kshitijgoel007 / auvforwarddynamics Goto Github PK

Objective: Make a PD controller for AUV to follow waypoints in the horizontal plane.

• Using two inputs : stern deflection and rudder deflection. Note : Bp and Bs are to be ignored as their control power is very low.
• Linearize the Yaw motion equation in the form x' = Ax + Bu and y = Cx + Du
• After the system is linearized, find the eigenvalues of the system.
• Using stability analysis, find the range of proportional and derivative gains.
• Determine final K_p and K_d values within this range.

attn : @shubhamjena @AlphaMale1996

Since the course-keeping controller is now tested against non-linear dynamics of NPS AUV II, next task is to incorporate long duration autonomy to course-keeping deep sea mission.

Example mission to be pursued is Vizag to Andaman - stretch of ~2000 kilometers. This would induce serious errors in the controller due to:

1. Oceanic currents
2. Earth's curvature

Depth has to maintained constant at 300 meters (so, error in Z has to be monitored too). Sway distance should be kept as minimum as possible.

We start working on this from 1/1/2017. Essential theory required to develop the stack will be circulated soon.

attn : @AlphaMale1996 @shubhamjena

Fix code organization in the commit 19755bb

attn : @shubhamjena

@AlphaMale1996 : remove redundant jpg and csv files, if they are not required anymore and recommit.

Current solver uses euler method. We need to switch to more accurate runge-kutta 4th order.

Task is to modify the solver file of forwarddynamics2.m to have a boolean on whether to use Euler or to use Runge-Kutta 4th order.

attn: @AlphaMale1996 @shubhamjena

The bot doesn't show up in proper shape when plotted. CG plots are now fine, so the next thing to be checked is rotation transform that is being applied to change the points that constitute the bot.

attn : @shubhamjena

Currently critical routines are not arranged in a modular way. Following modules (each separate MATLAB file) are intended:

• Euler Solver
• 6DOF equations - one file for NPS AUV II, one for kraken.
• File integrating solver and equations
• Force calculation and plotter. Component wise
• Moment calculation and plotter. Component wise
• Simulator files for different cases
• PD controller file.

attn: @AlphaMale1996 @shubhamjena

It's about time we start integrating Team AUV IIT Kharagpur's kraken 3.0 to our stack. For this first accurate equations are to be devised and accordingly a clone of forwarddynamics2.m has to be created for this case.

Phani Kiran will calculate the derivatives.

attn : @shubhamjena @AlphaMale1996

Integration of CUDA and Parallel Computing toolbox with MATLAB code.

Assigned : Kshitij Goel

The repository has to follow MATLAB style guidelines as mentioned here. Automatic documentation capability is really cool in MATLAB and we should get it generated quickly.

This can be open for KWoC.

attn : @shubhamjena @AlphaMale1996

Euler angles are diverging and that's why distances are being shown at arbitrary values.

attn : @AlphaMale1996