InnoFlight announces Cerberus Triple Redundant Autopilot Systems

The advanced Jupiter JM-1 multicopter autopilot is now fully redundant with the release of the Cerberus Triple Redundant Autopilot Module by InnoFlight.


Taoyuan, Taiwan (R.O.C) – April, 2016

InnoFlight (INF), the leading manufacturer of flight control systems, UAV and stabilised camera gimbals in Taiwan today announced the release of the world first triple redundant Cerberus multicopter autopilot system. The company has over ten years’ experience in design and development of flight stabilised system for VTOL type aircrafts as well as over thirty years combined experienced in UAV operations.

“We are very excited to announce the release of the Cerberus Triple Redundant autopilot system. Cerberus enable up to 5 standard Jupiter JM-1 flight controller and theirs built-in IMU’s, GPS, and Compasses to work together as a true triple redundant flight control system.

The Cerberus redundant autopilot system not only provides the obvious increase in safety but when comparing to a single Jupiter JM-1, our triple redundant system can process sensor data from all three Jupiter JM-1 thus greatly increase the accuracy and performance of the system.

The Cerberus enhanced redundant system not only enables redundancy among autopilot modules as a whole but also their internal sensors. As long as two Main Control Units and sensors from any module are functioning, the Cerberus would continue to provide normal flight control operations.

To understand the benefit of a true triple redundant system we can look at this simple example; if the reliability of a typical autopilot is 1/1000, then the reliability of the same autopilot in a triple redundant arrangement would be 1/1000 x 1/1000 = 1/1,000,000. If the same system is multiplied up to a quadruple redundant system, then the reliability will go up to 1/1 billion and so forth.

In fact, our Cerberus redundant system CAN bus technology can support up to 5 Jupiter JM-1 giving you unprecedented reliability never seen before.”

INF is expected to begin shipping the Cerberus in July of 2016. Customers can see the Cerberus fully functional prototype on display at NAB, and AUVSI this April/May.

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • @John Arne Birkeland @Thomas Butler   I agree with upstairs several, if the uav has  out of control, after all, is mostly human error, or flying in a bad environment, rarely because of flight control malfunction accident, so much the development of the redundant system is not much more meaningful.Too much instead of cable connection, reduce the safety factor, after all,less connection more stable, less in integration and multiple components to facilitate maintenance these two aspects to find the right balance point is feasible.

  • Same for me, unfortunately we are not allowed to fly anywhere outside of our club fields if we can't provide redundancy for the FC. As for me i would rather install a parachute and a killswitch as it gives me some sort of malfunction safety, i would also say if we are going more complex then chances of a failure are higher...

  • I just haven't seen the failure of sensors nor processors as a limiting factor in these systems.

    Just as in any electrical system, the inter-connecting system is usually the least reliable.  As far as my history with flying RC and multi-copters, as Rob intuitively points out above, my catastrophic failures have nearly all been pilot error.  ESC failures are my #1 hardware issue, followed by batteries, then motors. . . 

    My 2 centavos.

  • dual redundancy for pixhawk would also be great without an intermediate module...

    could be directly wired and the pwm signals are connected with a splitter. keep alive checks with a very relaxed voting system.

  • Still doesn't fix the "crap in, crap out" problem. Like the guy who sets his P rate to 1.2 and wonders why it keeps going out of control and crashing.

  • @James, yes, they have redundant sensors, but the sensor fusion in the EKF hasn't been as high performance as I'd like.  Up to about a year ago, the situation was actually that we doubled the chance of failure, because we were simply taking the average of the two sensors.  So if one went off... the whole thing was off.  The past year or so, things are much better, but until we truly have one EKF instance for each sensor, then it's not as good as it could be.

    Just the sort of complexities you only realize when people openly tell you how the systems work.

  • Speaking about doble GPS, I am wondering if it give us a real advantage. With a good setup, we do not have anymore GPS glitch, etc... (based on more 100 flights/week in auto missions). Is it worth the installation complexity, size, etc. Thank you for your advice. Max

  • James, I agree, though we do occasionally see individual sensor failures.  I think it makes sense to have redundancy at sensor level.

  • Surely it's an April Fools joke.

  • I think you guys misunderstood. We used 1/1000 as a random number to help explain redudancy.

    I know.  But the over-simplification doesn't help you make your case.  It also doesn't account for the fact that your "one in a billion control system" still passes through a single processor, which probably has the same failure rate as any single system.

This reply was deleted.