add warnings to jog_arm low_pass filter

[tylerjw]

Description

These warnings and errors handle warning the user of bad input parameter values for the low pass filter in jog arm. Here is the function of the feedback magnitude with respect to the low_pass_filter_coeff value: https://www.wolframalpha.com/input/?i=y+%3D+%281%2F%281%2Bx%29%29+*+%281-x%29

Here is a small python script that shows why these values are problematic:

def run_filter(mu, length):
  x0 = x1 = y1 = 0

  y = x = [0 for i in range(length)]
  x[0] = 1

  for i in range(length):
    x1 = x0
    x0 = x[i]

    y[i] = (1.0 / (1.0 + mu)) * (x1 + x0 + (1.0 - mu)*y1)
    y1 = y[i]

  print(mu, sum(y))

run_filter(2.0, 1000)
run_filter(1.0, 1000)
run_filter(0.5, 1000)
run_filter(0, 1000)
run_filter(-0.5, 1000)
run_filter(-0.999, 1000)
run_filter(-2, 1000)

output:

(2.0, 0.5000000000000002)
(1.0, 1.0)
(0.5, 1.9999999999999991)
(0, 1999.0)
(-0.5, inf)
(-0.999, inf)
(-2, nan)

Checklist

• Required by CI: Code is auto formatted using clang-format
• Extend the tutorials / documentation reference
• Create tests, which fail without this PR reference
• While waiting for someone to review your request, please help review another open pull request to support the maintainers

[codecov-io]

Codecov Report

Merging #1872 into master will increase coverage by 0.11%.
The diff coverage is 90.9%.

@@            Coverage Diff             @@
##           master    #1872      +/-   ##
==========================================
+ Coverage   50.15%   50.27%   +0.11%     
==========================================
  Files         313      313              
  Lines       24641    24628      -13     
==========================================
+ Hits        12359    12381      +22     
+ Misses      12282    12247      -35

Impacted Files	Coverage Δ
...xperimental/moveit_jog_arm/src/low_pass_filter.cpp	95.45% <90.9%> (-4.55%)	⬇️
...anning_scene_monitor/src/current_state_monitor.cpp	52.82% <0%> (-0.28%)	⬇️
...erimental/moveit_jog_arm/src/jog_cpp_interface.cpp	11.9% <0%> (ø)	⬆️
...veit_jog_arm/include/moveit_jog_arm/jog_arm_data.h	100% <0%> (ø)	⬆️
...cessing/src/time_optimal_trajectory_generation.cpp	76.2% <0%> (ø)	⬆️
...ics_plugin_loader/src/kinematics_plugin_loader.cpp	62.58% <0%> (+0.25%)	⬆️
moveit_core/robot_model/src/joint_model.cpp	71.42% <0%> (+6.66%)	⬆️
...veit_experimental/moveit_jog_arm/src/jog_calcs.cpp	72.79% <0%> (+8.1%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2395183...9bc7eee. Read the comment docs.

[jliukkonen]

Approved but see comments.

[AndyZe]

I think the error message about an unstable coefficient is awesome, greatly needed.

[tylerjw]

The last change was to make it C++11 compatible (the way I used static_assert was from C++17).

[henningkayser]

Isn't that a little paranoid? Also, what if someone initializes previous_measurements_ with another constant... ;)

[tylerjw]

I don't think it is too paranoid. The indexes of the array are explicitly used instead of loops below this and this check is just a compiler check. It won't add any code to the binary. This is just helping protecting against someone modifying this value.

I don't know why changing the values of previous_measurements_ with the reset method would affect this?

[henningkayser]

Sure, initialization using reset is not an issue. This check here only warns other developers from changing the value of FILTER_LENGTH in the header file. But you could just declare previous_measurement_ with another constant and this assertion will never be triggered, so it's not really a guarantee that logic doesn't break. I'm not concerned about performance, I just think that a simple comment in the header file is more effective than letting developers run into warnings at compile time.