The past few years I have been working on a project (Space Data Network) and selected flatbuffers as the transport. There were a few features that core flatbuffers didn't have that I needed, such as no mangling of names between different programming languages, per-field encryption while leaving the file id and length headers alone, ability to ingest JSON Schema in addition to IDL, and a few others.
Still a WIP but going well, all Apache 2.0, feedback welcome.