Combining UGV and UAV for Hostile Environment 

Unmanned systems has become a growing and common place within our armed forces. These systems have helped our forces for intelligence gathering, and to minimize human lose in the face of hostile environments and advisories. The call out for more sophisticated designs and operations for unmanned systems has been on the rise. UGVs have being used and developed to help complete the mission and again minimize human lose. "Systems are being developed by various defense agencies like DARPA’s LAGR program which focused on perception based off-road navigation in UGV’s. Foster-Miller’s TALON system is a remotely operated vehicle designed for missions ranging from combat to reconnaissance. They are working on incorporating virtual reality vision for their future vehicle called the Modular Advanced Armed Robotic System (MAARS). The US marines use their Gladiator Tactical Unmanned Ground Vehicle (TUGV) to minimize risks and eliminate threats during conflict. This vehicle is light weight and can be easily transported and deployed strategically for missions. The Indian Defense Research and Development Organization (DRDO) developed a fully automated UGV named Daksh for handling and destroying hazardous objects safely." (Sawarkar 2016). 

With the above technology on the rise, the need for additional inovations is called for. In Swarkar and his peers article they propose to combine both the Unmanned Ground Vehicle (UGV) with the Unmanned Aerial Vehicle (UAV) that is tele-operated by a human component for final decision making in the use of these systems purpose. The purpose of the system is that is can classify terrorist (or hostiles) by using a trained Deep Convolutional Neural Network called the Convolutional Neural Network (CNN) which is said to be trained to identify humans with weapons (i.e assault rifles, revolvers, etc.) The imaging which is fed back to the controller via a Head mount display (HDM) and virtual reality gives a probability that the "target" is hostile or not. By forming a green box around a "target" this would indicate that the "target" is not hostile if the box around the "target" is red then the probability that the "target" is hostile is increased. Though final decision is done by the tele-operated and not the unmanned system.

Why is this a combined system?  Is one of the main questions I asked while reading this article. The answer was a relatively simple one. If the UGV was damaged or could not reach a certain terrain then the UAV would come into affect. The UAV would be tele-operated but would use the dame Deep Convolutional Neural Network to identify both non hostile and hostile targets. Also in use of disaster situations, such as floods, landslides..etc. The UGV would be valuable to locate survivors on the ground, while the UAV could help to locate safe zones or be a beacon for helicopter rescue in certain areas. 

This design, though very inventive is still in the building process. The Deep Convolutional Neural Network at the moment only perceives those with a weapon as hostile. This at the moment will not work if they are escorting other armed servicemen as it will see them as a threat. This flaw also can not take into consideration the mindset that a hostile target may do in the event of a hostage being taken. If the hostile is quick thinking and knows the system that is being used they could fashion (though humorous to think about but has truth) a fake weapon and hand to the hostage. This would fool the system into identifying the hostage as a hostile and recommend action to be taken by the operator. This though would fall on the operator to not just take the systems advice but to also see and understand the situation before taking action.

This combination of a UGV and UAV is an interesting aspect of the building innovations and idea that are coming to Unmanned systems. As the technology continues to advance, I'm sure that the innovations will advance as well.

Reference: 

Sawarkar, A., Chaudhari, V., Chavan, R., Zope, V., Budale, A., & Kazi, F. (2016). HMD vision-based teleoperating UGV and UAV for hostile environment using deep learning.