Map with AI has been very helpful in the process of road mapping in Indonesia since last year. With the help of AI, it improved the quality and speed of mapping the roads. It guided mapper to visualize where the missing roads are.

Although in its implementation, remote mapping using AI assisted is not as easy as imagined. The mappers still need to give the roads classification manually which is difficult, because Indonesia has different road conditions in each island. I think this is clearly one of the reasons why human validation still plays an important role. One of the ways to face this challenge was to conduct field surveys on several different islands in Indonesia to gather local information. We have shared what we have got from the field with the Facebook team which I believed it could be used as valuable feedback in improving their tools.

Road data mapped with AI assistance were also useful in post-disaster mapping carried out by local communities. For example, post-earthquake and tsunami mapping activities in Palu, Central Sulawesi in September last year. Mapped roads helped navigate them to disaster affected areas, considering that many affected areas were located in remote areas where there were very limited information on road access. Mapped road networks combined with important infrastructure data (e.g health care, government offices, evacuation shelter, etc) that they have collected will be used to support the renewal of contingency plans and disaster management plans in the region.