Telescope Reviews: Dust Devils at Phoenix Landing Site

On an intial rub, it would seem that commanding a lateral movement shortly before touchdown would indeed introduce a huge extra set of unnecessary variables to the equation. But, when you fully analyze the dynamics of the landing sequence, in reality, it doesn't.

Irrelavent of the mode of approach, when the lander reaches the final phase, ie that of 'perched atop a retro-rocket', for the final deceleration and touchdown, it will have some form of lateral movement. During the final phase, the lateral movement MUST be stabilized to a point such that at touchdown, lateral forces will not be large enough to cause a rollover. No matter how you twist it, this is going to imply some sort of control required in the horizontal plane. Once you have created a system that has the required control in that plane to stabilize and minimize horizontal velocity, then there is sufficient control to also introduce a small lateral sidestep in the final stage. This is no more complex than reacting to a wind gust, something the lander MUST be capable of handling anyways.

I've never written the code to handle a vertical rocket descent, but, I have worked on aircraft autoland systems, so I have a rough idea of the concepts they would be working on. When you think of a vertical rocket descent in laymans terms, you think of 'stop horizontal movement, then descent to touchdown', that's a visual we get from cartoons as a kid, and it has stuck. In reality, the goal of a landing control module is to reach a specific set of parameters in the x/y/z planes. In this case, absolute x and y are irrelavent, not trying to hit a prepared runway, absolute z is important. Dx, Dy, and Dz are the other targets. The initial thought would be, Dx and Dy should be zero, Z should be very small, and Dz should be very small, then remove all propulsion and it'll just 'fall' the last inch or so. In reality, the final target will be a range of all those parameters, ie get the lander to an altitude and velocity from which it can 'fall' and still end up standing on the landing gear. To achieve the sidestep prior to touchdown, just change the horizontal target velocity from zero, to some number well within the capability of the lander to withstand rollover forces at impact.

I'll agree the lander is fraught with danger, but, it's not due to the sidestep on final approach. The bouncing beach ball landers had a HUGE range of acceptable values for Dx,Dy, and Dz at first impact. The vertical descent rocket has a very tiny range of acceptable values. The bouncing beach ball can also survive a lot of terrain issues by simply rolling over/around or bouncing off it. The vertical descent lander gets one chance, and, if it happens to come down on a steep slope or on top of a boulder, it's toast. The vertical descent lander doesn't have the benefit of real time feedback to adjust the landing zone, any visuals taken at the point of low approach wont reach earth till after the accident, so, no chance to re-direct it. For a historical and relavent reference, check out the final approach of Apollo 11 into the lunar touchdown. Without a real time 'change in plans' on the actual touchdown point, that lander would have been lost.

For the lander in question, vertical descent rocket as the final approach mode is the source of complexity. Changing the final horizontal target vector from zero to some small value at the end of the rocket descent adds no complexity to the system as a whole. What it does do, is change the range of acceptable velocity vectors, and, more importantly, the tolerance. Instead of 0/+3/-3 for a velocity target, they now have 1/+2/-4 as the acceptable range within which they can expect a successful touchdown. It is the job of the landing control module to hit this velocity at a pre-described altitude, with a specific vertical velocity range target, then shut down the propulsion and let gravity prevail. It's actually a relatively easy task, if you are confident that the underlying surface is flat, with a constant co-efficient of friction (that's why we build paved runways and helipads at airports). If the landing surface is unprepared, and has a slope, well, you pays your money, and takes your chances...