The Philippines SHARE Program in Aerial Imaging

Transcription

1 The Philippines SHARE Program in Aerial Imaging G. Tangonan, N. Libatique, C. Favila, J. Honrado, D. Solpico Ateneo Innovation Center This presentation is about our ongoing aerial imaging research in the Philippines. We present various aspects of our work such as the aircrafts we use, how we acquire and deliver our results to various stakeholders, and some use cases where this technology can be applied. This aerial imaging capability was developed by the Faculty and students of Ateneo de Manila University, working out of the Ateneo Innovation Center. Let me give you an overview of the capabilities we developed and the different uses cases we have demonstrated. We hope that this will provide you with a new capability for program management and productivity forecasting in Thailand. We are very proud to be a partner with Thailand in the SHARE program for several years now. Out of the spirit of sharing, we are grateful to our hosts, Dr. Anan and Dr. Asanee, for inviting us here today.

2 Recent Results Obtained in Three Regions of the Philippines In the last two years, have travelled extensively all over the country utilising aerial platforms to address various needs. We have engaged many partners in aquaculture, agriculture, and local government. The wide variety of interactions gives us important insight into how to optimise our equipment, field operations and post processing of imagery for different applications. Here we show our fixed wing drone being launched and the local collaborators we have that provide ground truth data for better understanding of imagery.

3 Present Inventory and Capabilities of Unmanned Aerial Vehicles models: both fixed wing and multirotor platforms powertrain: 5000 to 10000mAh lithium polymer batteries flight time: 15 to 25 min. - multirotor 45 to 60 min. - fixed wing resolution: 3 to 20 cm per pixel coverage (typical): 3 to 4 sq. km. at 10cm/pixel using fixed wing altitude: 100 to 500m payload: regular and modified consumer cameras Our aerial platforms include both fixed wing and rotary UAVs. We are mainly using customised platforms, which we engineer ourselves, that are optimised for the specific applications like agriculture and disaster risk reduction. Shown on the right side are other operational characteristics describing our field units. Most of the work we describe here are for fixed wing aircraft. The craft are battery powered, flight times are 45 minutes to one hour. We fly at altitudes of 100 m to 500 m and the attainable resolution ranges from 3 to 20 cm. Typical flights can cover 3-4 square kilometers in these times. The aerial imaging crew is four to five technical staff.

4 Flight Planning: Sample Mission area of interest parameters computed by software based on user input Autopilot: 3DR APM 2.6 Software: Mission Planner Payload: Canon EOSM w/ 22mm lens Altitude: 400m Overlap: 80% Sidelap: 65% So how do we get the data? Shown here is a sample mission plotted in Mission Planner, the software we use in the field. The red box is the area we want to map and we tell the software other parameters such as the model of the camera we re using, altitude, and desired overlap and sidelap. It then generates the waypoints indicated by the green pins and computes the necessary flight parameters such as trigger time and distance between legs. every shot of the camera anticipates the next stage of image stitching, so we use high overlap between images to minimize the chances of missing areas. We program autopilot software so the UAV executes this Flight Plan automatically. Shown below are some important flight parameters that are crucial to successful mission planning and execution.

5 Lake Resource Management San Pablo, Laguna : Monitoring Fishkills, Aquaculture Output calculating density and tracking ownership of fishpens in Lake Palakpakin and many more DOST- AIC Aerial Imaging Consortium Here is an example from the SHARE project in which we show an aerial image of a crater lake, Lake Palapakin in San Pablo, Laguna. This lake is covered with fish cages where the fisherfold raise tilapia. We have developed a completed lake management system which links the aerial imagery with sensors on the health. In this way we can monitor the health of the lake, the development in the area, and effects of weather and climate change in lake areas. It shows the power of combining aerial imagery with ground truth data taken from instruments and the local stakeholders. This is done in the same spirit that your excellent program in CyberBrain for agriculture/aquaculture.

6 Web- based Decision- support System Web- based decision support system A B C DO Conductivity We integrate all the data gathered by the sensor network, by the UAV and by the stakeholders into a web decision- support application. In the design, the user can view the aerial maps of the lake area taken at a particular date. Data that can be overlaid on the map including, but not limited to, bathymetry data gathered by lake researchers, fish pen trackers to identify their registrations and floating field servers deployed in different locations. We can also display the 24/7 sensor data from each field server when a particular field server is selected. The user can also view the photos and notes taken by the stakeholders at a particular date. With this decision- support system, the stakeholders can interact with each other in making plans for the lake, crafting policies and mitigating impacts of disasters on the lake. This system can help counteract the complex problems in the management of the lake resource. Recently we completed the aerial imaging of all the Seven Lakes in San Pablo Laguna. We are now in the process of putting deploying field servers in all the Lakes.

7 Final Product - Delivery of Aerial Images distributed through: plain printouts map layouts with GIS software soft copies for uploading to websites online viewing platform Photo by PhilRice After the mission, the photos acquired are stored and fed to a stitching software to generate aerial maps of the study sites. We distribute it to our partners and stakeholders through printed and electronic means. The resulting data from the stitching software can stretch from a few hundred megabytes to several gigabytes depending on the mission characteristics. This can make it hard for some users to view it in their personal workstations. With that in mind, we have developed an online delivery platform which we ll demo on the next video slide.

8 Online Data Delivery Platform Here, the user can select which area he or she would like to view. Under a location, it can also contain several maps based on when the mission was taken. There are also lightweight GIS tools which the user can utilize to place markers and other information related to the site. The user can append the database with useful information on the database. The video shows how planners and forecasters can add ground truth data to the information system. In this way we can go from image taking to user interaction in a span of a few days.

9 Mangrove Reforestation Management with UAVs Aerial Surveys for Determining Mangrove Reforestation Areas Let us now give you some actual use cases as examples. We hope you see that in Thailand very similar use cases can be studied and analyzed, using the same technology developed locally by Universities like Kasetstart. In this example we were contracted by the World Food Programme to map the coastlines and river tributaries in Aparri, Cagayan to help state universities and local government units in their mangrove reforestation efforts. Where there were mangroves rice fields are protected, without the mangroves much of the rice crop was being lost to flooding. After receiving the images, planners were able to draw in the yellow shaped above as priority areas where mangroves needed to be planted. This is a state of the art tool for Mangrove Reforestation programs using aerial imaging.

10 Precision Agriculture with Near IR and RGB Cameras dual camera setup (Canon S100) VIS NIR We ve had many partners that are involved with agriculture research and there is increasing interest in the use of multispectral cameras to monitor their field plantations. The chlorophyl in healthy plants reflect near infrared light strongly during the different growth stages. One popular parameter is the Normalised Difference Vegetation Index which measures the relationship between visual and NIR reflectance. To this end we developed our own low cost near infrared imaging platform by modifying consumer grade cameras. The infrared blocking filter was removed and a filter was attached to get the NIR signal. The Near IR camera and a regular RGB camera are flown by the UAV to get multispectral data. Figure A shows satellite imagery that covers around 2 square kilometers. Many agriculture programs rely on satellite images exclusively, but you notice that the satellite imagery has very low resolution - tens of meters at best for available sources. In contrast you our high resolution images in Figures B to D greatly complement and enhance the image taken in the particular area. We can distinguish different crops and the different stage of development of the crops with high resolution. This data can be compared to ground truth data taken from individual farms, together they can give good estimates of future crop yield.

11 Sample Dual Camera Output: Rice Field Mapping VIS NIR *study conducted in partnership with IBM and WWF Philippines Shown here is a very sample output of our multispectral platform taken with a team of researchers from the WWF for Nature Philippines. These are taken in a 1000 hectare area in Isabela Province, composed mainly of rice fields. It shows the regular visual map on the left and a single band, greyscale image on the right for the NIR.

12 NDVI Maps for Monitoring Crop Performance Ground and Aerial Sensor Scatter Plot Histogram NDVI classification map based on NDVI values *study conducted in partnership with IBM and WWF Philippines Given the two maps, we can compute for the NDVI and see the prevalent growth stage during the mission. Our measurements were also correlated with ground NDVI measurements deployed by our partners. In this particular example, we can see a big part are in the newly planted stage, signified by the brown patches, and another major part is in the vegetative stage, shown by the green shapes. It allows us to see the distribution of these growth stages and allows us to have yield estimates given a particular planting season.

13 Use Case: Disaster Damage Assessment baseline map structure classification damage extent mapping *ground truth data and figures c/o our partners in Catholic Relief Services We were also asked to map a coastline after typhoon Haiyan for damage assessment. Our partners used our maps as baseline to distribute with their ground teams to classify structures and to assess extent of the damage per household.

14 Use Case: Infrastructure Monitoring farm-to-market road planning and progress monitoring Here is a unique use case that deals with the development of farm to market roads in the Philippines. Illustrated here is the use of UAVs in community planning in Javier, Leyte. The Mayor of Javier designed and proposed a farm to market road, using aerial images taken by our team. He was able to design the placement of a bridge over a river that had not been mapped before. He was able to even count the number of trees that had to be uprooted during construction. He was able to make right of way decisions, while designing the road. Our maps were valuable in design of the road and bridges in 2012, we returned in 2013 and 2014 to document from the air the progress of this major local program. Here is pretty clear evidence of the potential for aerial imaging in government programs that involve public private partnerships in infrastructure development.

15 Thank you very much for your kind attention. We are very happy to have been invited to this important meeting. Thank you for your kind attention.