Remotely Operated Vehicles (ROVs) are everywhere – autonomous rovers, vision-guided robots, self-driven cars and machine-operated industrial arms and remote-navigated war-tanks. What sets these apart is the absence of a human driver and the control of the vehicle taken over by a video-assisted navigation system & driving system. There are two types of ROVs:
- Vehicles with on-board video analytics: Live feed from one/more camera(s) is processed/analyzed to identify road-signs, obstacles, people etc. and the driving controls of the vehicle are automatically managed by an embedded device on the vehicle.
- Given the processing requirements for real-time video analytics, such an embedded device requires a high-performance processor. Power consumption for such a device may be a consideration for rovers and robots.
- Enhanced systems like Lane Departure Warning, Parking Assistant, Cross Traffic Warning, Traffic Sign Recognition, Rear-Collision Warning and Blind Spot Detection can further improve the chances of an accident-free drive.
- Vehicles with remote control: The live feed from one/more camera(s) is encoded real-time and streamed to a remote location over wireless network. Analysis of the video feed is done by a human or a machine at this remote location. Commands related to driving controls are then related back to the vehicle.
- Given the time-critical nature of the task, the round-trip latency (including processing latency on the device & remote location, and network latency) needs to be very low. Latency as low as 40-50msec is essential because this matches the typical reaction time for a human driver. Lower latencies are even better and desirable.
- This is more typical for ROV operation over unfamiliar terrain or surroundings. Applications include hazardous material disposal, industrial robots, reconnaissance drones or espionage tanks.
If you are looking at developing such a product, here are 6 easy tips to get started:
- Assess the mission criticality of your application and determine whether you need on-board analytics.
- Know how much power you can afford for the vehicle’s navigation system to consume.
- Based on 1 and 2, determine the embedded platform that meets your needs. For example, Ittiam’s neonCaster hardware boards (Clove, Magna) allow live capture of up to 2 Full-HD cameras at the same time, besides video encoding, recording & streaming.
- If opting for on-board analytics, select efficient realizations of algorithms that do what you need and offer configurability to adapt to your usage scenario. Couple these analytics with the right pre-processing algorithms (ex: Contrast Enhancement, De-noising, High Dynamic Range or HDR) for more accurate results. Ittiam offers a wide variety of software IPs to meet these needs.
- If opting for remote control of the vehicle, low latency video streaming is critical. Ittiam’s adroitLive Media SDK, for example, can achieve as low as 40-50 msec glass-to-glass latency for Full-HD (1080p60) video.
- If opting for remote control of the vehicle, ensure that you can provide the best video streaming experience over wireless. Adapting to variable channel conditions is key here. Ittiam’s wireless streaming technology provides the best trade-offs between latency, quality and fluctuating network bandwidth.
Hope this gets you all ready to roll.
Get. Set. Go….