#第二届立创大赛#大疆精灵2开源头追与osd抬头显示系统 - 立创电子设计大赛

主页 >

立创电赛

赛事

第二届立创电子设计大赛

版本协议

BSD

描述

一、作品简介 大疆精灵2是大疆出的精灵系列第二代四轴飞行器，发布于2015年，现在大疆公司不断推陈出新，大疆精灵系列已经出到了第4代，大疆精灵2集成度没有后边的几代集成度高，但相对的，留给DIYER们改造的空间也相对较大，本制作就是针对大疆精灵2的can信息进行解码，来扩展改造。此制作分为两大部分。 第一是针对需要对大疆精灵2系列四轴飞行器的电池信息，GPS信息等通过stm32加can芯片，进行CAN接口解码，叠加到摄像头图像里，方便了解飞行器飞行时的状态，并解码出两个遥控器通道进行云吞俯仰的操作，配合地面遥控器头追，实时控制云吞的俯仰状态，并通过模拟图传实时发送到地面飞手的VR眼镜里，让飞手有一种身临其境的飞行快感。 第二是头追的制作，头追传感器采用ardunio与GY-85九轴模块设计，放置在VR眼镜上部，采集飞手的头部动作，通过遥控器发送信号给四轴飞行器，从而控制飞行器摄像头的俯仰与航向，让飞手犹如坐在飞行器机仓般的感觉。 <div> </div> <img src="//image.lceda.cn/pullimage/NyyVcJKJ4qWajpQiomrMdnpRKCp0RB5tQNAq541S.jpeg" alt="" width="501" height="680"> <img src="//image.lceda.cn/pullimage/gVU0ed4AdGZVoU6iG8TOT7N7GmcXABXB6Thc4kXY.jpeg" alt="" width="527" height="673"><img title="点击查看大图" src="//" alt=""> <img src="//image.lceda.cn/pullimage/6ObQ6DHaduPFuQvztNCMtFFptP4hcYz2pp2eVgMc.jpeg" alt="" width="520" height="679"><img title="点击查看大图" src="//" alt=""> <img title="点击查看大图" src="//" alt=""> 二、系统构架图 OSD系统框图 <img src="//image.lceda.cn/pullimage/r1Qp0mBO9TGmFlP8iK19Kicny3KWLGm4rdDot8Hb.jpeg" alt="" width="564" height="365"><img title="点击查看大图" src="//" alt=""> 头追系统框图<img src="//image.lceda.cn/pullimage/YHo4l9h6yruOWpobaYpEyjCdlMiIctvWWJu3ONMc.jpeg" alt="" width="516" height="276"> <img title="点击查看大图" src="//" alt=""> 三、硬件部分的描述 OSD原理图与成品图 <img src="//image.lceda.cn/pullimage/oRXkKIP3YnO4fRMPSqTAbbdCS97z84phoHaS3d2A.jpeg" alt="" width="892" height="495"> <img src="//image.lceda.cn/pullimage/Vx82wSMvyP7GZHXVl0YGkrar6nXMWH3PJrhNzB3l.jpeg" alt="" width="815" height="609"> <img title="点击查看大图" src="//" alt=""> 头追系统成品图：<img src="//image.lceda.cn/pullimage/vZixqCJ2kFSd1EDtBWZv458Hc5AzQNiU0BQOqLJD.jpeg" alt="" width="507" height="676"> <img title="点击查看大图" src="//" alt=""> 主要原理： 1.osd 抬头显示系统。 大疆精灵2飞控系统，会不断的向CAN总线传送各种飞控状态信息和控制信息，改OSD设计就是解码精灵2的can数据，解析出四轴飞行器的电池信息，GPS信息，飞行高度，和遥控器摇杆数值，各种飞控信息通过stm32的can模块接受并解码，送osd芯片结合摄像头传来的图像信息，进行字符叠加，再通过5.8G视频无线图传模块传送到地面接收。地面通过5.8g视频接收模块接收并显示图像。解码出的遥控器的两个通道的数据，由stm32转换成pwm输出给舵机，从而控制舵机的转向和俯仰，实现挂在舵机上的摄像头随地面遥控信号动作。 2.头追系统。 头追系统主要由一块ardunio 板和GY-85 9轴传感器和两个I2C接口的DAC芯片组成，这里采用了现成的模块搭接。ardinuo 读取9轴传感器的状态和磁力计的数据，进行融合计算，转换成俯仰，横滚，朝向的变化数值，并通过DAC芯片转换成电压给遥控器的两个通道读取，模拟遥控器的电阻器动作，从而实现头部的动作模拟成摇杆的动作，控制飞行器上的舵机转动。这样飞行器上摄像头就随着地面控制人员的头部运动而运动，模拟飞机上上下左右看的感觉。 <a class="ke-insertfile" href="http://club.szlcsc.com/article/downFile_D1B31AB6F874CC95.html" target="_blank">skylink_osd_p2_V1.0_PCB.zip</a> (下载次数：1052) 原理图和PCB文件（采用kicad软件) 四、材料清单（BOM列表) stm32f103c8t6 商品编号：C8734 tja1050 商品编号：C112947 at7456 商品编号：C82351 atmega328 商品编号：C14877 mcp4725 商品编号：C61423 GY-85模块 https://item.taobao.com/item.htm?spm=a230r.1.14.36.76bf5236XgcH1&id=35070081530&ns=1&abbucket=5#detail 五、软件部分的描述（选填） OSD软件流程 <img src="//image.lceda.cn/pullimage/lh4ZmsgnEfCVqHzevSqvx5qyFIqAkSJjVVPE8Qsz.jpeg" alt="" width="417" height="345"><img title="点击查看大图" src="//" alt=""> 头追软件流程 <img src="//image.lceda.cn/pullimage/iz54HpyVOxBZh48uLUTgbR34zX6CWyXsGqjsEtIr.jpeg" alt="" width="470" height="329"><img title="点击查看大图" src="//" alt=""> osd 大疆can信息解码代码 <pre class="prettyprint lang-cpp">uint16_t NazaCanDecoderLib::decode() { uint16_t msgId = NAZA_MESSAGE_NONE; if(can.available()) { can.read(&RxMessage); if(RxMessage.StdId == 0x090) canMsgIdIdx = 0; else if(RxMessage.StdId == 0x108) canMsgIdIdx = 1; else if(RxMessage.StdId == 0x7F8) canMsgIdIdx = 2; else return msgId; // we don't care about other CAN messages for(uint8_t i = 0; i < RxMessage.DLC; i++) { canMsgByte = RxMessage.Data[i]; if(collectData[canMsgIdIdx] == 1) { msgBuf[canMsgIdIdx].bytes[msgIdx[canMsgIdIdx]] = canMsgByte; if(msgIdx[canMsgIdIdx] == 3) { msgLen[canMsgIdIdx] = msgBuf[canMsgIdIdx].header.len; } msgIdx[canMsgIdIdx] += 1; if((msgIdx[canMsgIdIdx] > (msgLen[canMsgIdIdx] + 8)) || (msgIdx[canMsgIdIdx] > 256)) collectData[canMsgIdIdx] = 0; } // Look fo header if(canMsgByte == 0x55) { if(header[canMsgIdIdx] == 0) header[canMsgIdIdx] = 1; else if(header[canMsgIdIdx] == 2) header[canMsgIdIdx] = 3; else header[canMsgIdIdx] = 0;} else if(canMsgByte == 0xAA) { if(header[canMsgIdIdx] == 1) header[canMsgIdIdx] = 2; else if(header[canMsgIdIdx] == 3) { header[canMsgIdIdx] = 0; collectData[canMsgIdIdx] = 1; msgIdx[canMsgIdIdx] = 0; } else header[canMsgIdIdx] = 0;} else header[canMsgIdIdx] = 0; // Look fo footer if(canMsgByte == 0x66) { if(footer[canMsgIdIdx] == 0) footer[canMsgIdIdx] = 1; else if(footer[canMsgIdIdx] == 2) footer[canMsgIdIdx] = 3; else footer[canMsgIdIdx] = 0;} else if(canMsgByte == 0xCC) { if(footer[canMsgIdIdx] == 1) footer[canMsgIdIdx] = 2; else if(footer[canMsgIdIdx] == 3) { footer[canMsgIdIdx] = 0; if(collectData[canMsgIdIdx] != 0) collectData[canMsgIdIdx] = 2; } else footer[canMsgIdIdx] = 0;} else footer[canMsgIdIdx] = 0; if(collectData[canMsgIdIdx] == 2) { if(msgIdx[canMsgIdIdx] == (msgLen[canMsgIdIdx] + 8)) { if(msgBuf[canMsgIdIdx].header.id == NAZA_MESSAGE_MSG1002) { float magCalX = msgBuf[canMsgIdIdx].msg1002.magCalX; float magCalY = msgBuf[canMsgIdIdx].msg1002.magCalY; headingNc = -atan2(magCalY, magCalX) / M_PI * 180.0; if(headingNc < 0) headingNc += 360.0; float headCompX = msgBuf[canMsgIdIdx].msg1002.headCompX; float headCompY = msgBuf[canMsgIdIdx].msg1002.headCompY; heading = atan2(headCompY, headCompX) / M_PI * 180.0; if(heading < 0) heading += 360.0; sat = msgBuf[canMsgIdIdx].msg1002.numSat; gpsAlt = msgBuf[canMsgIdIdx].msg1002.altGps; lat = msgBuf[canMsgIdIdx].msg1002.lat / M_PI * 180.0; lon = msgBuf[canMsgIdIdx].msg1002.lon / M_PI * 180.0; alt = msgBuf[canMsgIdIdx].msg1002.altBaro; float nVel = msgBuf[canMsgIdIdx].msg1002.northVelocity; float eVel = msgBuf[canMsgIdIdx].msg1002.eastVelocity; spd = sqrt(nVel * nVel + eVel * eVel); cog = atan2(eVel, nVel) / M_PI * 180; if(cog < 0) cog += 360.0; vsi = -msgBuf[canMsgIdIdx].msg1002.downVelocity; msgId = NAZA_MESSAGE_MSG1002; } else if(msgBuf[canMsgIdIdx].header.id == NAZA_MESSAGE_MSG1003) { uint32_t dateTime = msgBuf[canMsgIdIdx].msg1003.dateTime; second = dateTime & 0x3f; dateTime >>= 6; //0b00111111 minute = dateTime & 0x3f; dateTime >>= 6; //0b00111111 hour = dateTime & 0x0f; dateTime >>= 4; //0b00001111 day = dateTime & 0x3f; dateTime >>= 5; if(hour > 7) day++; //0b00011111 month = dateTime & 0x0f; dateTime >>= 4; //0b00001111 year = dateTime & 0x7f; //0b01111111 gpsVsi = -msgBuf[canMsgIdIdx].msg1003.downVelocity; vdop = (double)msgBuf[canMsgIdIdx].msg1003.vdop / 100; double ndop = (double)msgBuf[canMsgIdIdx].msg1003.ndop / 100; double edop = (double)msgBuf[canMsgIdIdx].msg1003.edop / 100; hdop = sqrt(ndop * ndop + edop * edop); uint8_t fixType = msgBuf[canMsgIdIdx].msg1003.fixType; uint8_t fixFlags = msgBuf[canMsgIdIdx].msg1003.fixStatus; switch(fixType) { case 2 : fix = FIX_2D; break; case 3 : fix = FIX_3D; break; default: fix = NO_FIX; break; } if((fix != NO_FIX) && (fixFlags & 0x02)) fix = FIX_DGPS; msgId = NAZA_MESSAGE_MSG1003; } else if(msgBuf[canMsgIdIdx].header.id == NAZA_MESSAGE_MSG1009) { for(uint8_t j = 0; j < 10; j++) { rcIn[j] = msgBuf[canMsgIdIdx].msg1009.rcIn[j]; } #ifndef GET_SMART_BATTERY_DATA battery = msgBuf[canMsgIdIdx].msg1009.batVolt; #endif // rollRad = msgBuf[canMsgIdIdx].msg1009.roll; // pitchRad = msgBuf[canMsgIdIdx].msg1009.pitch; rollRad = msgBuf[canMsgIdIdx].msg1009.stabRollIn; pitchRad = msgBuf[canMsgIdIdx].msg1009.stabPitchIn; roll = (int8_t)(rollRad*0.1); pitch = (int8_t)(pitchRad*0.1); // roll = (int8_t)(rollRad * 180.0 / M_PI); // pitch = (int8_t)(pitchRad * 180.0 / M_PI); mode = (NazaCanDecoderLib::mode_t)msgBuf[canMsgIdIdx].msg1009.flightMode; msgId = NAZA_MESSAGE_MSG1009; } #ifdef GET_SMART_BATTERY_DATA else if(msgBuf[canMsgIdIdx].header.id == NAZA_MESSAGE_MSG0926) { battery = msgBuf[canMsgIdIdx].msg0926.voltage; batteryPercent = msgBuf[canMsgIdIdx].msg0926.chargePercent; for(uint8_t j = 0; j < 3; j++) { batteryCell[j] = msgBuf[canMsgIdIdx].msg0926.cellVoltage[j]; } msgId = NAZA_MESSAGE_MSG0926; } #endif } collectData[canMsgIdIdx] = 0; } } } return msgId; }</pre> 头追系统融合代码 <pre class="prettyprint lang-cpp">//-------------------------------------------------------------------------------------- // Func: Filter // Desc: Filters / merges sensor data. //-------------------------------------------------------------------------------------- void FilterSensorData() { int temp = 0; // Used to set initial values. if (resetValues == 1) { #if FATSHARK_HT_MODULE digitalWrite(BUZZER, HIGH); #endif resetValues = 0; tiltStart = 0; panStart = 0; rollStart = 0; UpdateSensors(); GyroCalc(); AccelCalc(); MagCalc(); panAngle = 0; tiltStart = accAngle[0]; panStart = magAngle[2]; rollStart = accAngle[1]; #if FATSHARK_HT_MODULE digitalWrite(BUZZER, LOW); #endif } // Simple FilterSensorData, uses mainly gyro-data, but uses accelerometer to compensate for drift rollAngle = (rollAngle + ((gyroRaw[0] - gyroOff[0]) * cos((tiltAngle - 90) / 57.3) + (gyroRaw[2] - gyroOff[2]) * sin((tiltAngle - 90) / 57.3)) / (SAMPLERATE * SCALING_FACTOR)) * gyroWeightTiltRoll + accAngle[1] * (1 - gyroWeightTiltRoll); tiltAngle = (tiltAngle + ((gyroRaw[1] - gyroOff[1]) * cos((rollAngle - 90) / 57.3) + (gyroRaw[2] - gyroOff[2]) * sin((rollAngle - 90) / 57.3) * -1) / (SAMPLERATE * SCALING_FACTOR)) * gyroWeightTiltRoll + accAngle[0] * (1 - gyroWeightTiltRoll); panAngle = (panAngle + ((gyroRaw[2] - gyroOff[2]) * cos((tiltAngle - 90) / 57.3) + (((gyroRaw[0] - gyroOff[0]) * -1) * (sin((tiltAngle - 90) / 57.3)) ) + ( ((gyroRaw[1] - gyroOff[1]) * 1) * (sin((rollAngle - 90) / 57.3)))) / (SAMPLERATE * SCALING_FACTOR)) * GyroWeightPan + magAngle[2] * (1 - GyroWeightPan); if (TrackerStarted) { // All low-pass filters tiltAngleLP = tiltAngle * tiltRollBeta + (1 - tiltRollBeta) * lastTiltAngle; lastTiltAngle = tiltAngleLP; rollAngleLP = rollAngle * tiltRollBeta + (1 - tiltRollBeta) * lastRollAngle; lastRollAngle = rollAngleLP; panAngleLP = panAngle * panBeta + (1 - panBeta) * lastPanAngle; lastPanAngle = panAngleLP; float panAngleTemp = panAngleLP * panInverse * panFactor; if ( (panAngleTemp > -panMinPulse) && (panAngleTemp < panMaxPulse) ) { temp = servoPanCenter + panAngleTemp; channel_value[htChannels[0]] = (int)temp; } float tiltAngleTemp = (tiltAngleLP - tiltStart) * tiltInverse * tiltFactor; if ( (tiltAngleTemp > -tiltMinPulse) && (tiltAngleTemp < tiltMaxPulse) ) { temp = servoTiltCenter + tiltAngleTemp; channel_value[htChannels[1]] = temp; } float rollAngleTemp = (rollAngleLP - rollStart) * rollInverse * rollFactor; if ( (rollAngleTemp > -rollMinPulse) && (rollAngleTemp < rollMaxPulse) ) { temp = servoRollCenter + rollAngleTemp; channel_value[htChannels[2]] = temp; } } }</pre> <a class="ke-insertfile" href="http://club.szlcsc.com/article/downFile_8264D38F29A88364.html" target="_blank">ebox_skylink_osd_p2_fw_v1.0.zip</a> (下载次数：1648) osd源代码 <a class="ke-insertfile" href="http://club.szlcsc.com/article/downFile_3583DB278302C59B.html" target="_blank">skylink_osd_p2_PC_v1.0.zip</a> (下载次数：1031) osd上位机代码 <a class="ke-insertfile" href="http://club.szlcsc.com/article/downFile_92A62F71F1CC038A.html" target="_blank">DIY_Headtracker.zip</a> (下载次数：1775) 头追代码与设计 六、作品演示 七、总结 这个方案是模拟遥控器的摇杆动作的，所以要接线到遥控器的两个通道，这样其实很别扭，下一步是要用433m无线模块替代模拟摇杆，做到真正的免连线头追。 <h3>更多项目详情见链接：http://club.szlcsc.com/article/details_7034_1.html 本项目归立创社区“diylink”所有</h3>