Why can't robotic vehicles gain traction for the ground force?

Thanks to Paul and Joe for the link!

via Defense One.

A research team from Stamford, Conn. has developed an amphibious drone that they are currently testing with the Marines. The GuardBot is a robot ball that swims over water at about 4 miles per hour and then rolls along the beach, at as much as a 30-degree incline and 20 miles per hour.
It uses a nine-axis stabilization, “pendulum motion” propulsion system, which moves the bot forward by shifting the center of gravity back and forth and a variety of steering algorithms.
It took creator Peter Muhlrad some seven years to develop, but now that it’s complete Muhlrad says it can be rapidly produced in various sizes. Company documents suggest it can be scaled down to units as small as 10 cm and as large as nine feet. The company is planning to develop a prototype that’s 6 feet in diameter.

And then this from Military.com.

After years of being featured at trade shows and trotted out for high-ranking Marine Corpsofficials, the Marines' barrel-chested Legged Squad Support System -- known affectionately as the robotic mule -- has been put out to pasture.
The machine, which resembles a headless pack mule made of metal, came about through a $32 million, two-and-a-half year contract between the Pentagon's research arm, known as the Defense Advanced Research Projects Agency, and Google Inc.'s Boston Dynamics, of Waltham, Massachusetts.
DARPA teamed up with the Marine Corps Warfighting Lab to create an autonomous four-legged creature that could lighten troops' load by carrying 400 or more pounds of weight, according to reports about the 2010 contract.
A second contract worth almost $10 million was awarded in 2013 for an additional phase of the LS3 program that would demonstrate how the legged robot would work by following troops on foot through rugged terrain, carrying their gear, and interpreting verbal and visual commands. The contract also provided for the construction of an enhanced version of LS3 that featured a quieter power supply and better survivability against small arms fire.
In 2012, Marine Corps Commandant Gen. James Amos attended a demonstration of the prototype's capabilities at Joint Base Myer-Henderson Hall, Virginia. At the time, Amos expressed pride in the developing technology and said it was getting close to something the Marines might use, according to reports.

Full disclosure.  I think both of the above projects are steaming piles of dung.  I don't think they're worth the money and the goals were/are questionable at best.  Any idiot could tell you that the "legged squad support system" wasn't going to work and should have been done away with long ago.  But that's just a small part of the story when it comes to robotic vehicles for the ground force.

Remember this from BAE, the Black Knight?

What about the Oshkosh Terramax Heavy Logistics Vehicle?

And my final example, this blast from the past from Carnegie Melon Crusher Robotic Combat Vehicle?

We've been pushing the stone for a long time now, but it has yet to bear fruit...but the weird thing is that we have developed vehicles to fill the goals set out but haven't bitten.

The cynic in me says its because of a lack of funds.  But some of these designs go back to when the Pentagon was throwing money around like wives at a swingers party.  The critic in me says that its cheaper to simply man armored vehicles and payout the insurance in case someone is killed or pay for healthcare if they're injured.

Either way it doesn't matter.  We have capable designs to fulfill the logistics, scouting and even heavy combat roles now.  We just aren't buying and instead we're heading towards flights of fancy.