Crash on Quadplane Simulation

[RMIShane]

Issue details

I'm using Vagrant SITL to do quadplane simulation.

Launching with this command: sim_vehicle.py -L KSFO -j 2 -v ArduPlane -f quadplane

then uploading an auto mission: wp load ../Tools/autotest/ArduPlane-Missions/KSFO-VTOL.txt

then: arm throttle

then: auto

At this point the simulation crashes.

auto
FBWA> APM: Executing nav command ID #84
APM: Resetting previous waypoint
Got MAVLink msg: COMMAND_ACK {command : 11, result : 0}
APM: Reset alt target to -12.6
waypoint 1
/vagrant/Tools/autotest/run_in_terminal_window.sh: line 54: 27548 Aborted                 (core dumped) $cmd $* &> "$filename" < /dev/null

Version

ArduPlane 3.8beta5 (git Tag ArduPlane-beta)

Logs

Started model quadplane at 37.619373,-122.376637,5.3,118 at speed 1.0
Starting sketch 'ArduPlane'
Starting SITL input
Using Irlock at port : 9005
bind port 5760 for 0
Serial port 0 on TCP port 5760
Waiting for connection ....
Loaded defaults from /vagrant/Tools/autotest/default_params/quadplane.parm
bind port 5762 for 2
Serial port 2 on TCP port 5762
bind port 5763 for 3
Serial port 3 on TCP port 5763
Smoothing reset at 0.001
Loaded defaults from /vagrant/Tools/autotest/default_params/quadplane.parm
ERROR: Floating point exception - aborting

[peterbarker]

On Sun, 14 May 2017, RMIShane wrote:

I'm unable to replicate.

ArduPlane 3.8beta5 (git Tag ArduPlane-beta)

Please provide a specific git hash.

ERROR: Floating point exception - aborting

That's very bad.

[RMIShane]

commit 92d505598bd58507a9b618e33f3d849a664c6143

Pulled everything again today and still having the same issue.

(seams to just crash with quadplane) following these steps:

1. vagrant ssh -c "sim_vehicle.py -L KSFO -j 2 -v ArduPlane -f quadplane --console --map"

2. wp load ../Tools/autotest/ArduPlane-Missions/KSFO-VTOL.txt

3. arm throttle

4. AUTO

[image: Inline image 1]

On Mon, May 15, 2017 at 11:25 PM, Peter Barker notifications@github.com wrote:

On Sun, 14 May 2017, RMIShane wrote:

I'm unable to replicate.

ArduPlane 3.8beta5 (git Tag ArduPlane-beta)

Please provide a specific git hash.

ERROR: Floating point exception - aborting

That's very bad.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/ArduPilot/ardupilot/issues/6245#issuecomment-301665156, or mute the thread https://github.com/notifications/unsubscribe-auth/AbXHWsO7nHxOVAknvsLwly1dO6bIOXcgks5r6RcZgaJpZM4NaswV .

[RMIShane]

Got MAVLink msg: MISSION_ACK {target_system : 255, target_component : 0, type : 0, mission_type : 0} APM: Flight plan received arm throttle FBWA> APM: Throttle armed Got MAVLink msg: COMMAND_ACK {command : 400, result : 0} ARMED auto FBWA> APM: Executing nav command ID #84 APM: Resetting previous waypoint Got MAVLink msg: COMMAND_ACK {command : 11, result : 0} APM: Reset alt target to 16.8 waypoint 1 /vagrant/Tools/autotest/run_in_terminal_window.sh: line 54: 26969 Aborted (core dumped) $cmd $* &> "$filename" < /dev/null FBWA> no link link 1 down no link no link no link no link

On Tue, May 16, 2017 at 12:03 AM, Shane Roberts rmi.shane@gmail.com wrote:

commit 92d505598bd58507a9b618e33f3d849a664c6143

Pulled everything again today and still having the same issue.

(seams to just crash with quadplane) following these steps:

1. vagrant ssh -c "sim_vehicle.py -L KSFO -j 2 -v ArduPlane -f quadplane --console --map"

2. wp load ../Tools/autotest/ArduPlane-Missions/KSFO-VTOL.txt

3. arm throttle

4. AUTO

[image: Inline image 1]

On Mon, May 15, 2017 at 11:25 PM, Peter Barker notifications@github.com wrote:

On Sun, 14 May 2017, RMIShane wrote:

I'm unable to replicate.

ArduPlane 3.8beta5 (git Tag ArduPlane-beta)

Please provide a specific git hash.

ERROR: Floating point exception - aborting

That's very bad.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/ArduPilot/ardupilot/issues/6245#issuecomment-301665156, or mute the thread https://github.com/notifications/unsubscribe-auth/AbXHWsO7nHxOVAknvsLwly1dO6bIOXcgks5r6RcZgaJpZM4NaswV .

[peterbarker]

On Mon, 15 May 2017, RMIShane wrote:

commit 92d505598bd58507a9b618e33f3d849a664c6143

Pulled everything again today and still having the same issue.

(seams to just crash with quadplane) following these steps:

1. vagrant ssh -c "sim_vehicle.py -L KSFO -j 2 -v ArduPlane -f quadplane --console --map"

2. wp load ../Tools/autotest/ArduPlane-Missions/KSFO-VTOL.txt

3. arm throttle

4. AUTO

Replicated, thanks! Any time someone can replicate something in the vagrant VM makes me happy.

[peterbarker]

Program received signal SIGFPE, Arithmetic exception.
__ieee754_expf_sse2 () at ../sysdeps/i386/i686/fpu/multiarch/e_expf-sse2.S:116
116     ../sysdeps/i386/i686/fpu/multiarch/e_expf-sse2.S: No such file or directory.
(gdb) bt
#0  __ieee754_expf_sse2 ()
    at ../sysdeps/i386/i686/fpu/multiarch/e_expf-sse2.S:116
#1  0xb7e145c8 in __GI___expf (x=180.639633)
    at ../sysdeps/ieee754/flt-32/w_expf.c:26
#2  0x08127eb6 in std::exp (__x=180.639633) at /usr/include/c++/6/cmath:246
#3  0x081283a2 in SITL::Plane::liftCoeff (this=0x820bff0, alpha=3.14159274)
    at ../../libraries/SITL/SIM_Plane.cpp:77
#4  0x081289d3 in SITL::Plane::getForce (this=0x820bff0, 
    inputAileron=5.66840718e-05, inputElevator=-0.0838166773, 
    inputRudder=0.000192956984) at ../../libraries/SITL/SIM_Plane.cpp:192
#5  0x08129003 in SITL::Plane::calculate_forces (this=0x820bff0, input=..., 
    rot_accel=..., body_accel=...) at ../../libraries/SITL/SIM_Plane.cpp:280
#6  0x081295de in SITL::QuadPlane::update (this=0x820bff0, input=...)
    at ../../libraries/SITL/SIM_QuadPlane.cpp:102
#7  0x08114a61 in HALSITL::SITL_State::_fdm_input_local (
    this=0x81d2e60 <sitlState>)
    at ../../libraries/AP_HAL_SITL/SITL_State.cpp:275
#8  0x0811418e in HALSITL::SITL_State::_fdm_input_step (
    this=0x81d2e60 <sitlState>)
    at ../../libraries/AP_HAL_SITL/SITL_State.cpp:137
#9  0x0811452b in HALSITL::SITL_State::wait_clock (this=0x81d2e60 <sitlState>, 
    wait_time_usec=45383771) at ../../libraries/AP_HAL_SITL/SITL_State.cpp:190
#10 0x0811579c in HALSITL::Scheduler::delay_microseconds (
    this=0x81d53a4 <sitlScheduler>, usec=2764)
    at ../../libraries/AP_HAL_SITL/Scheduler.cpp:45
#11 0x0811569a in AP_HAL::Scheduler::delay_microseconds_boost (
    this=0x81d53a4 <sitlScheduler>, us=2764)
    at ../../libraries/AP_HAL/Scheduler.h:30
#12 0x080a0d37 in AP_InertialSensor::wait_for_sample (
    this=0x81cf140 <plane+3360>)
    at ../../libraries/AP_InertialSensor/AP_InertialSensor.cpp:1302
#13 0x0804acf1 in Plane::loop (this=0x81ce420 <plane>)
    at ../../ArduPlane/ArduPlane.cpp:119
#14 0x081133ed in HAL_SITL::run (this=0x81d56a0 <AP_HAL::get_HAL()::hal>, 
    argc=11, argv=0xbffff454, callbacks=0x81ce420 <plane>)
    at ../../libraries/AP_HAL_SITL/HAL_SITL_Class.cpp:88
#15 0x0804d59e in main (argc=11, argv=0xbffff454)
    at ../../ArduPlane/ArduPlane.cpp:1060
(gdb) print alpha
No symbol "alpha" in current context.
(gdb) up
#1  0xb7e145c8 in __GI___expf (x=180.639633)
    at ../sysdeps/ieee754/flt-32/w_expf.c:26
26      ../sysdeps/ieee754/flt-32/w_expf.c: No such file or directory.
(gdb) 
#2  0x08127eb6 in std::exp (__x=180.639633) at /usr/include/c++/6/cmath:246
246       { return __builtin_expf(__x); }
(gdb) 
#3  0x081283a2 in SITL::Plane::liftCoeff (this=0x820bff0, alpha=3.14159274)
    at ../../libraries/SITL/SIM_Plane.cpp:77
77              double sigmoid = ( 1+exp(-M*(alpha-alpha0))+exp(M*(alpha+alpha0)) ) / (1+exp(-M*(alpha-alpha0))) / (1+exp(M*(alpha+alpha0)));
(gdb) print alpha
$1 = 3.14159274

That's an interesting value for alpha....

[tridge]

looks like we need a constrain_float() in there

[peterbarker]

@RMIShane If you felt like diving into this one, you can run SITL with --debug --gdb and trace through as I did above. Did you want to do that?

A quick workaround would be to change the Vagrant virtual machine to be a 64-bit instance instead of a 32-bit instance.

[OXINARF]

This should have been fixed by https://github.com/ArduPilot/ardupilot/commit/85ebe923b663a9c7e2cc367fe303c007eeaace07

[RMIShane]

Awesome!

Thank you!

On Fri, Sep 15, 2017 at 8:39 AM Francisco Ferreira notifications@github.com wrote:

Closed #6245 https://github.com/ArduPilot/ardupilot/issues/6245.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ArduPilot/ardupilot/issues/6245#event-1251327325, or mute the thread https://github.com/notifications/unsubscribe-auth/AbXHWh8BdoZBhTO2CstKxK1A0wqe-YI8ks5sipo1gaJpZM4NaswV .