Beagle 2 PAW and ARM

Source: Bridges et al. 2017, Royal Society Open Science, figure 4 [5]; CC BY 4.0.

The problem

A Mars lander arm has to put instruments and sampling tools on rock and soil it has seen only in stereo images, with no human able to correct a placement in real time. A UHF relay pass gives a few contacts per sol, so every arm sequence is planned a sol ahead and executed blind [8]. Contact with an unseen surface cannot be rehearsed, so the arm either carries a force sensor or detects contact by another means and stops. The soil is not what orbital data suggest: Viking measured drift material at 1,200 kg/m3 with 0.7 to 3.0 kPa cohesion, and crusty or cloddy material behaves differently under a tool [13]. InSight later found a duricrust at the HP3 site with penetration resistance of 0.5 to 1.2 MPa and cohesion of 4 to 25 kPa [18].

Beagle 2 added a mass constraint: the lander was 33 kg on a 69 kg probe limit, and the program aimed at the highest ratio of payload to support systems flown to Mars, which meant an integrated design and minimal or zero redundancy [9]. ESA’s mission description makes the same point: the volume was set by the aeroshell, and no supplier makes boxes at that scale [14]. The consequence for the arm was an architecture in which the instruments sit on the wrist, so the arm exists to point a payload.

What has flown

Beagle 2 ARM and PAW (never exercised). The Anthropomorphic Robotic Manipulator has five degrees of freedom and 109 cm of reach [9]. It ends in the Position Adjustable Workbench, a fixed cluster of stereo cameras, microscope, X-ray spectrometer, Mossbauer spectrometer, rock corer grinder, wind sensor, torch, spoon and the launch tube for the PLUTO mole [9]. The lander reached Isidis Planitia on 25 December 2003 and never communicated [11]. HiRISE imaging identified it in 2014 and 2015, and the object is consistent with lid deployment and partial solar panel deployment, which would have left the UHF antenna covered [5]. The ARM, PAW and PLUTO were therefore never operated on Mars, and every number for them is an as-built or as-tested value [9].

Mars Surveyor 2001 and Phoenix arms. The Mars Surveyor 2001 lander arm was a 4 degree of freedom back-hoe design with graphite epoxy links, canceled before flight [1]. Its MVACS-class lineage produced the Phoenix arm and, refurbished, the InSight arm [2]. The Phoenix arm, flown in 2008, is a 2.4 m back-hoe with a scoop, dual blades and a rasp, built to acquire dry and icy soil for the deck instruments [3]. Over more than 150 sols it executed 53 dig sequences across 10 trenching areas and delivered 18 samples to instruments [4].

InSight Instrument Deployment Arm (2018 to 2022). The arm has 4 degrees of freedom and 1.8 m reach [15]. It placed the wind and thermal shield over the seismometer to within 1.5 cm of the intended position, the first robotic stacking of one instrument over another [15]. When the HP3 mole stalled, the arm took part in an anomaly recovery campaign that included pinning the mole with the scoop, which gained 5 cm of penetration, and pushing on the mole back cap [16]. Scoop pushes and indentation tests near the pit were reduced to soil cohesion of 3 to 14.5 kPa and friction angle of 30 to 34 degrees [17].

Perseverance arm (2021). The arm has 5 actuators over two titanium links and a redundant 6 degree of freedom force-torque sensor at the wrist [21]. Resolver angles on sol 11 agreed with pre-launch values within 5 mrad. The first surface contact came 2 mm early at about 30 N, and docking to the bit carousel, the first on another planet, succeeded on the first attempt [21]. Of the force-corrected dockings on Mars analyzed, 9 of 9 succeeded [22]. Sensor-level calibration reached 3 sigma residuals below 67 N and 7.5 Nm, and the two-stage system calibration met the 65 N and 16 Nm absolute accuracy requirement [23].

Beagle 2 design

With the instruments on a workbench at the end of the arm [14], every instrument placement is an arm placement, and the requirements on the ARM are optical rather than mechanical. Digital elevation model construction needs two or three perspectives with targets in focus, the multispectral panorama needs the PAW pointed straight up and rotated, and the close-up lens standoff constrains orientation [9]. The piezoelectric washer is quiet during gentle motion, so it works as a contact event detector and not as a force sensor. Advance toward a target proceeds until contact or a timeout, after which the sequence aborts to a safe position [9].

JPL’s Planetary Dexterous Manipulators task took a different path with ultrasonic actuation and camera-guided autonomous control of rover arms [6]. A lander replica at JPL had already demonstrated vision-based positioning with a two meter gas-deployed telescopic arm [10].

The stereo camera system is a pair of CCD micro-cameras with 48 degree optics and a 209 mm baseline, each behind a twelve-position filter wheel, and one position holds the close-up lens giving 60 micrometer per pixel at 80 mm [7]. The wide angle mirror is a deployment safety device: it enters the right camera’s field of view after the lid opens, so planners can unstow the PAW without collision [9]. The rock corer grinder removes weathering rind, then cores the ground patch by combined hammering and rotation [9].

PLUTO is a tethered mole derived from a Russian self-burying penetrometer, driven by an internal inertial hammer so that it needs no reaction force from the lander after initial penetration [9]. Communication was UHF only, through antennas in the lid [8].

What is being tried now

Force sensing and contact handling. Beagle 2 detected contact with a piezoelectric load washer used as an event trigger and then drove on for a short interval [9]. The Perseverance arm measures six-axis force and moment and verified it in flight, with better than 37 percent margin to a plus or minus 65 N and plus or minus 16 N m accuracy requirement [21]. Both approaches have flown.

Arm safety and motion planning. Operations teams automate the check that the arm will not hit terrain or load an unstable rover. Curiosity ran the Slip Risk Assessment Process 1258 times with a 93.6 percent go rate, falling to 71.4 percent at parking tilts above 25 degrees [24]. Proximeter, a JPL tool, computes worst-case clearance between the arm model and a stereo terrain mesh in under 50 ms per placement [25]. A sampling-based planner for Perseverance arm sequencing cut mean viapoints from 3.86 to 1.08, because each viapoint is an arm command a rover planner reviews by hand [26]. The slip-risk process and Proximeter are operational tools on flown missions; the planner is a published method for them.

Autonomous science targeting. PIXL on Perseverance re-bins each spectrum into 22 pseudo-intensities and extends the dwell on points matching rules trained on Earth, adding 30 s to a 10 s standard dwell [27]. AEGIS has selected SuperCam targets autonomously, reaching 170 targets in flight with no rover-level safety protections triggered [28]. Both have flown. Causal-curiosity reinforcement learning for classifying unknown objects with a planetary manipulator is a simulation study only [29].

Pneumatic sampling. PlanetVac replaces scoop and arm delivery with a gas jet. On Blue Ghost Mission 1 at Mare Crisium in March 2025 it collected 6.5 volume units of regolith in a 5 second acquisition [30][31]. This has flown.

Rotary-percussive drills. TRIDENT is a 1 m class drill of 26.5 kg with 2 J blows, built for PRIME-1 and VIPER [32]. In field campaigns at Haughton Crater it drilled 7.80 m in total, 2.4 m of it into ice or ice-cemented breccia, and five fault states were observed and recovered [33]. An Antarctic campaign of 15 boreholes showed that drill telemetry alone can detect the depth of ice-rock and frozen-ground boundaries [34]. ExoMars carries a drill with a 2 m depth goal that has not flown [35]. These are ground-tested for Mars or lunar use.

Autonomous excavation and sampling. The Europa Lander sampling autonomy campaign ran 58 field tests on the Matanuska Glacier [36]. The Mars Sample Return arm testbed Bradbury is a 7 degree of freedom, 2.3 m arm built to reproduce the flight arm’s loads [37]. OWLAT, a lander-manipulator testbed, tests autonomy for ocean world landers [38]. These are ground demonstrations.

Cold-capable arms. COLDArm is a 4 degree of freedom, 2 m arm with bulk metallic glass gear actuators for lunar operations, tested in thermal, dust seal and regolith simulant campaigns [39]. It has been ground tested.

Soil properties from the arm. Perseverance’s abrading bit has been used on Earth as a pressure-sinkage and shear instrument for fitting soil strength parameters [40]. Microgravity penetrometry has been run in parabolic flight in support of the MMX sampler, with Phobos simulants [41]. Both are ground results.

Mole-type penetrators. PLUTO was the first self-penetrating mole designed for a lander arm, and it never ran on Mars [9]. The InSight HP3 mole used the same hammering principle and reached only 37 cm tip depth against a 3 to 5 m requirement, because a cohesive duricrust denied the hull friction the hammer recoil needed [12]. The nominal recoil is 5.4 N [12]. The flight model is 400 mm long and 0.93 kg [19]. HP3 still measured soil temperature to about 36 cm [20]. The hammer principle has flown once and did not reach depth.

The numbers

Parameter	Value
ARM degrees of freedom	5, PAW permanently attached at the wrist
ARM length, fully extended	109 cm
Joint actuator	DC brushed motor through a 100:1 harmonic gearbox
Typical joint speed	0.6 deg/s, axes 1 to 3
Contact sensing	piezoelectric load washer, 3 pC/N, stop delay about 100 ms
ARM mass	2.11 kg
PAW mass	2.75 kg
Rock corer grinder	348 g, 6 W, grinding up to 6 mm of rind, 10 mm patch
PLUTO	890 g, 3 W, vertical sampling depth 1.5 m baselined, about 0.20 cm3 sample
Landed mass	33.18 kg, of which payload 8.896 kg

Beagle 2 values from [9].

Arm	Year	Reach and degrees of freedom	Source
Phoenix	2008	2.4 m, back-hoe	[3]
InSight IDA	2018	1.8 m, 4	[15]
Perseverance	2021	2 m class, 5	[21]
COLDArm	ground test	2 m, 4	[39]
Beagle 2	lost 2003	1.09 m, 5	[9]

Event	Result	Source
InSight shield placement	within 1.5 cm	[15]
HP3 mole tip depth	37 cm	[12]
Curiosity slip-risk go rate	93.6 percent	[24]
Perseverance force-corrected dockings	9 of 9	[22]
TRIDENT at Haughton Crater	7.80 m total drilled	[33]
Phoenix dig sequences	53	[4]

What is not established

Beagle 2 never communicated, so no ARM, PAW or PLUTO performance exists from Mars [11]. Every specification above is an as-designed or as-tested value from the hardware program, not an as-landed measurement [9]. The identification in orbital imagery rests on visual and morphological interpretation of an object of order 1.5 m at 25 to 30 cm per pixel, with no in situ confirmation; the candidate rear cover and main parachute are inconclusive, and the partial panel deployment is inferred from the image [5]. The design-stage mass and diameter in the ESA mission description are pre-launch values [14].

The PLUTO hammering principle was never tested against Martian soil, and the HP3 outcome does not transfer to it directly [12]. Beagle 2 published no accuracy budget comparable to the Perseverance arm calibration [23]. The InSight shield placement is a single measurement, not a demonstrated repeatability [15].