[feature] DW1000 Driver, 802.15.4 Frame & Ranging by Maju-Ketchup · Pull Request #361 · roboterclubaachen/xpcc

Maju-Ketchup · 2018-05-30T09:14:28Z

Added driver for DW1000 with generic and STM32F4_disco examples
Added MAC Frame for IEEE 802.15.4 with unitttests
Added Ranging algorithms for UWB modules with unittests

It is one year since I started this project of including a DW1000 Ultra-Wide-Band Radio and I finally want you to merge it into the dev branch, so everyone can profit from it.
Everything is on discussion names, paths etc...

The code is tested and used at my workplace and the bugs are mostly fixed.
I see forward to an useful discussion and I know this merge request will take some time.
Please be aware that I do the public changes in my free-time of which I do not have much.

rleh · 2018-05-30T10:00:20Z

+{
+	setdelayedtrxtime(time);
+	int result;
+	result  = rxenable(1);


CircleCI fails with

src/xpcc/driver/radio/dw1000/dw1000_impl.hpp:853:22: error: no matching function for call to 'xpcc::Dw1000<Spi, Cs, Reset, Irq>::rxenable(int)' src/xpcc/driver/radio/dw1000/dw1000.hpp:434:2: note: candidate: static int xpcc::Dw1000<Spi, Cs, Reset, Irq>::rxenable(xpcc::dw1000::RX_Mode)

This line should probably read as follows

result = rxenable(RX_Mode::START_RX_DELAYED);

(RX_Mode::START_RX_DELAYED == 1)

You are right. I changed using INTs to a custom enum some time ago. Seems to be a leftover.

rleh · 2018-05-30T13:04:31Z

The atmega-unittests test in CircleCI fails because there is not ostream-operator for long long types for AVRs.

xpcc/src/xpcc/io/iostream.hpp

Lines 512 to 515 in 2c898ae

    
           #if not defined(XPCC__CPU_AVR) 
        
                   void 
        
                   writeUnsignedLongLong(unsigned long long unsignedValue, uint_fast8_t base, size_t width, char fill, bool isNegative); 
        
           #endif

What is the reason for that? @salkinium

salkinium · 2018-05-30T14:07:19Z

What is the reason for that?

Perfomance concerns on AVRs. The issue was that this will make ALL printf calls use 64-bit integer math, regardless of whether you need it or not.
#315

We can selectably enable this with a macro for AVR targets that want to use this. By default it would be limited to 32-bit integers, but if you really need this, you can add a such a define flag to the build and enable this. In modm this would then be enabled with a module option.

Maju-Ketchup · 2018-05-30T14:56:19Z

I will disable unittests with 64bit uint for AVRs for now. If they work fine on STM32 and/or Linux it should be enough...

When this is done you can finally critique my additions.

rleh · 2018-06-09T20:40:45Z

I've split @Maju-Ketchup commits to get a better overview.

And replaced execfile() by +exec(compile(open(xpccpath + '/scons/SConstruct', "rb").read(), xpccpath + '/scons/SConstruct', 'exec')) in some SConstruct files.

rleh

Some examples are double in examples/stm32f4_discovery/ and examples/generic/, but differ slightly.

No file from the examples has a copyright header.

rleh · 2018-06-09T20:48:01Z

+	{
+		LedBlue::set();
+
+		//--------------------------------------- Start sending init ---------------------------------------------------------------


Unnecessarily long line.
(Not only here)

rleh

Should positioning/ be located in math/? @salkinium

rleh · 2018-06-09T20:54:38Z

@@ -0,0 +1,17 @@
+// coding: utf-8
+/* Copyright (c) 2018, Roboterclub Aachen e.V.


rleh · 2018-06-09T20:54:46Z

@@ -0,0 +1,199 @@
+/**
+*  Copyright (c) 2018, Marten Junga (Github.com/Maju-Ketchup)


rleh · 2018-06-09T20:55:15Z

+						   1,anchor0[1]*anchor0[1],anchor1[1]*anchor1[1],anchor2[1]*anchor2[1],
+						   1,anchor0[2]*anchor0[2],anchor1[2]*anchor1[2],anchor2[2]*anchor2[2]};
+
+//Matrix<floatunit, 3,4> A0(a0);


rleh · 2018-06-09T20:59:25Z

+struct multilateration
+{
+public:
+	/*!


Indentation from here to L140 looks funky 😉

rleh · 2018-06-09T21:03:13Z

+private:
+
+
+	/**


Empty documentation comments have no advantage

rleh · 2018-06-09T21:09:11Z

+		buffer[0] = buffer[0];
+		return true;}
+	static bool sendAt(int length,uint16_t buffer[], uint32_t time) {
+		int a = length;


What is the intention?

Maybe:

(void) length; (void) buffer; (void) time;

rleh · 2018-06-09T21:11:29Z

@@ -0,0 +1,21 @@
+// coding: utf-8
+/* Copyright (c) 2009, Roboterclub Aachen e.V.


rleh · 2018-06-09T21:12:45Z

+*
+*/
+
+#ifndef XPCC_FRAME802154_CPP


Remove L15-L18

rleh · 2018-06-09T21:17:33Z

+struct frame802154{
+public:
+	enum
+			Header : uint8_t


Indentation

rleh · 2018-06-09T21:18:15Z

+private:
+	void setbeginvalues();
+
+	//	void setbeginvalues();


salkinium

Should positioning/ be located in math/?

Yes in xpcc/math/positioning with the top-level header being xpcc/math/positioning.hpp.

salkinium · 2018-06-09T21:38:29Z

+* license. See the file `LICENSE` for the full license governing this code.
+*
+*
+* The headder contains the class implementation of the IEEE standart 802.15.4-2011 Frame


rleh · 2018-06-09T21:22:44Z

@@ -0,0 +1,1341 @@
+/*<------------------------------------------------------------------------------------------------------------------


Should this file be located in ext/? @salkinium

~~Yes, anything that's not of our original creation should go to ext, this would go into ext/decawave/deca_regs.hpp?~~

~~What license is this file? "All rights reserved." means your not allowed to do anything with it. What file did you adapt this from?~~

Edit: This file is an original creation not by Decawave.

rleh · 2018-06-09T21:23:20Z

+
+#include "dw1000.hpp"
+
+//--------------------------------------------------------------XPCC-Functions-Puplic------------------------------------------------------------------------


No need for the dashes

rleh · 2018-06-09T21:24:55Z

+//-----------------------------------------
+
+
+//------------------------------------------------------- CLASS BEGIN -----------------------------------------------------------------------


rleh · 2018-06-09T21:25:10Z

+namespace xpcc
+{
+//-----Without those at this place there will be unfixable bugs (cannot put them into params.hpp) -  if anyone knows why pls change
+//-----------------------------------------


rleh · 2018-06-09T21:26:05Z

+#include "./params.hpp"
+namespace xpcc
+{
+//-----Without those at this place there will be unfixable bugs (cannot put them into params.hpp) -  if anyone knows why pls change


rleh · 2018-06-09T21:34:07Z

+
+
+//Lot of Parts copied out of the DWM1000 API added some 'easier to use functions' *mja*
+//"Don't include this file directly, use 'dw1000.hpp' instead!"


#ifndef _DW1000_ #error "Don't include this file directly, use 'dw1000.hpp' instead!" #endif #include dw1000.hpp // optional, some IDEs need this for better code highlighting or completition

rleh · 2018-06-09T21:34:55Z

+
+template < typename Spi, typename Cs, typename Reset, typename Irq >
+uint32_t
+xpcc::Dw1000< Spi, Cs, Reset, Irq >::TX_ANT_DLY = 16436;


Where does this magic value come from?

rleh · 2018-06-09T21:36:16Z

+
+template < typename Spi, typename Cs, typename Reset, typename Irq >
+xpcc::dw1000::IRQreason
+xpcc::Dw1000< Spi, Cs, Reset, Irq >::getIRQReason() // TODO


rleh · 2018-06-09T21:47:55Z

+	rxreset();
+}
+
+//--------------------------------------------------------DWM1000-API----------------------------------------------------------------------------------


rleh · 2018-06-09T21:50:00Z

+int
+xpcc::Dw1000< Spi, Cs, Reset, Irq >::starttx(TX_Mode mode)
+{
+	/*int retval = SUCCESS ;


Please delete

salkinium · 2018-06-09T21:56:35Z

@@ -0,0 +1,1341 @@
+/*<------------------------------------------------------------------------------------------------------------------


~~Yes, anything that's not of our original creation should go to ext, this would go into ext/decawave/deca_regs.hpp?~~

~~What license is this file? "All rights reserved." means your not allowed to do anything with it. What file did you adapt this from?~~

Edit: This file is an original creation not by Decawave.

salkinium · 2018-06-09T21:57:33Z

+#ifndef _DECA_REGS_H_
+#define _DECA_REGS_H_
+
+#include <xpcc/architecture/platform.hpp>


This doesn't depend on platform, only on stdint.h.

My predecessor took this file from the original Decawave API, which is a horrible C code

salkinium · 2018-06-09T22:00:59Z

+static constexpr uint32_t DEV_ID_REV_MASK         =0x0000000FUL;    /*<Revision */
+static constexpr uint32_t DEV_ID_VER_MASK         =0x000000F0UL;    /*<Version */
+static constexpr uint32_t DEV_ID_MODEL_MASK       =0x0000FF00UL;    /*<The MODEL identifies the device. The DW1000 is device type =0x01 */
+static constexpr uint32_t DEV_ID_RIDTAG_MASK      =0xFFFF0000UL;    /*<Register Identification Tag =0xDECA */


Was the original file done with a lot of #defines? And if so why did you translate this to C++ constexpr?

Historical reasons my predecessor did that. I don't know why

edit - I think the reasoning about it was to be focused on types, that an error using the wrong kind of type could be found faster. But I'm just guessing here.
I also changed some (at least 2) of the registers vs the original file because they were wrong...

Ok, then this is original work, since you cannot copyright the data, only the formatting. Can you put your predecessor and yourself as copyright owners (BSD-licensed)?

I would then not put this into ext/, since we also don't put the register definitions of our sensor devices from modm/driver into ext/ either.

salkinium · 2018-06-24T19:06:22Z

FYI: I've just put up our deprecation notice for xpcc, as we are currently moving on to modm as its successor. I will be porting all new code to modm for the next 6 months, but in Jan 2019 I will be archiving this repository and then no more contribution can make it into xpcc.

So… no pressure, but just keep it in mind for your time planning.

salkinium · 2018-06-24T19:20:37Z

I already asked you this before in #318 (comment), but are you/your employer still ok with dual licensing this contribution as BSD for xpcc and MPLv2 for modm?

salkinium · 2018-09-02T03:38:33Z

@Maju-Ketchup any update on this? We still want this in xpcc, tell us how we can help you with this.

Maju-Ketchup · 2018-09-04T13:54:24Z

Hi there,

sorry for my long absence but i had no time due to exams and a high workload.
My idea for this is that I split my work into 3 pieces as @rleh commented.

-The Decawave DW1000 driver
-The MAC-Frame of IEEE 802.15.4
-The trilateration and ranging algorithms/framework

The last part is highly work in progress and throws some issues.
So i would leave it out for now.

my workprocess would be the following for the next step:

-Close this merge request.
-Add the 802.15.4 Frame to xpcc
-Add the Decawave driver to xpcc
-Add the Ranging Framework to xpcc/modm
-Add the examples

I will add feature branches for everything during the next week.
Or is there something else on your mind?

@salkinium
We are totally okay at my workplace to change the license to MPLv2 and include it into modm.

salkinium · 2018-09-04T15:07:37Z

Or is there something else on your mind?

No, that sounds perfectly fine for me. This doesn't need to be perfect, just good enought that it'll pass the examples/tests, so I can port it to modm with reasonable assurance that I didn't break anything.
Although the API changes will only be cosmetic, and not functional.

salkinium · 2019-01-03T19:03:36Z

As announced, I'm going to archive the xpcc repository soon. You fork will not be influenced, you can still use it as long as you need to. Looking at your code, I'm fairly positive that it can be ported to modm with only a few changes (and a healthy dose of s/xpcc/modm), if you still want to do that.

rleh reviewed May 30, 2018

View reviewed changes

Maju-Ketchup force-pushed the feature/DecaWaveDWM1000Driver branch 4 times, most recently from 2fb42f4 to 097cc98 Compare May 30, 2018 11:12

Maju-Ketchup force-pushed the feature/DecaWaveDWM1000Driver branch from 097cc98 to 45ea976 Compare June 4, 2018 07:45

Marten Junga added 6 commits June 9, 2018 22:12

[examples] STM32F4 Discovery: move NRF24 examples to subdir

3f28c3f

[examples] STM32F4 Discovery: add dw1000 examples

077c67d

[examples] generic: add dw1000 examples

bf1308b

[positioning] add multilateration and ranging

f94ae76

[communication] add Frame802154

9262656

[driver] add Decawave DW1000 driver

935ca59

rleh force-pushed the feature/DecaWaveDWM1000Driver branch from 45ea976 to 935ca59 Compare June 9, 2018 20:39

rleh suggested changes Jun 9, 2018

View reviewed changes

rleh reviewed Jun 9, 2018

View reviewed changes

salkinium reviewed Jun 9, 2018

View reviewed changes

rleh reviewed Jun 9, 2018

View reviewed changes

rleh added feature help wanted communication example drivers labels Jun 9, 2018

rleh added this to the 2018q2 milestone Jun 9, 2018

salkinium reviewed Jun 9, 2018

View reviewed changes

salkinium removed this from the 2018q2 milestone Jun 28, 2018

strongly-typed mentioned this pull request Mar 29, 2020

Share your modm projects modm-io/modm#360

Closed

		@@ -0,0 +1,17 @@
		// coding: utf-8
		/* Copyright (c) 2018, Roboterclub Aachen e.V.

		@@ -0,0 +1,199 @@
		/**
		* Copyright (c) 2018, Marten Junga (Github.com/Maju-Ketchup)

		@@ -0,0 +1,21 @@
		// coding: utf-8
		/* Copyright (c) 2009, Roboterclub Aachen e.V.

		@@ -0,0 +1,1341 @@
		/*<------------------------------------------------------------------------------------------------------------------


		#include "dw1000.hpp"

		//--------------------------------------------------------------XPCC-Functions-Puplic------------------------------------------------------------------------



		//Lot of Parts copied out of the DWM1000 API added some 'easier to use functions' mja
		//"Don't include this file directly, use 'dw1000.hpp' instead!"

		private:


		/**

Conversation

Maju-Ketchup commented May 30, 2018

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rleh commented May 30, 2018

Uh oh!

salkinium commented May 30, 2018

Uh oh!

Maju-Ketchup commented May 30, 2018

Uh oh!

rleh commented Jun 9, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rleh left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rleh left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

salkinium left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

salkinium Jun 9, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

salkinium Jun 9, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rleh commented Jun 9, 2018 •

edited

Loading

rleh left a comment •

edited

Loading

salkinium Jun 9, 2018 •

edited

Loading

salkinium Jun 9, 2018 •

edited

Loading

Maju-Ketchup Jun 11, 2018 •

edited

Loading

salkinium commented Jun 24, 2018 •

edited

Loading